Vascular endothelial growth factor in diabetes induced early retinal abnormalities

Diabetes Research and Clinical Practice 65 (2004) 197–208

Vascular endothelial growth factor in diabetes inducedearly retinal abnormalities

Mark Cukiernika, Denise Hileetoa, Terry Evansa, Suranjana Mukherjeea,Donal Downeyb, Subrata Chakrabartia,∗

a Department of Pathology, University of Western Ontario, London, Ont., Canada N6A 5C1b Department of Nuclear Medicine and Diagnostic Radiology, University of Western Ontario, London, Ont., Canada N6A 5C1

Received in revised form 17 November 2003; accepted 2 February 2004

Abstract

Increased vascular permeability and blood flow alterations are characteristic features of diabetic retinal microangiopathy.The present study investigated vascular endothelial growth factor (VEGF) and its interactions with endothelin (ET) 1 and 3,endothelial, and inducible nitric oxide synthase (eNOS, iNOS) in mediating diabetes induced retinal vascular dysfunction. MaleSprague Dawley rats with streptozotocin (STZ) induced diabetes, with or without VEGF receptor signal inhibitor SU5416 treat-ment (high or low dose) were investigated after 4 weeks of follow-up. Colour Doppler ultrasound of the ophthalmic/centralretinal artery, retinal tissue analysis with competitive RT-PCR and microvascular permeability were studied. Diabetes causedincreased microvascular permeability along with increased VEGF mRNA expression. Increased vascular permeability wasprevented by SU5416 treatment. Diabetic animals showed higher resistivity index (RI), indicative of vasoconstriction with in-creased ET-1 and ET-3 mRNA expression, whereas eNOS and iNOS mRNA expressions were un-affected. SU5416 treatmentcorrected increased RI via increased iNOS in spite of increased ET-1, ET-3 and VEGF mRNA expression. Cell culture (HU-VEC) studies indicate that in part, an SU5416 induced iNOS upregulation may be mediated though a MAP kinase signallingpathway. The present data suggest VEGF is important in mediating both vasoconstriction and permeability in the retina in earlydiabetes.© 2004 Elsevier Ireland Ltd. All rights reserved.

Keywords:Diabetes; Blood flow; VEGF; SU5416; Permeability

1. Introduction

The major target of diabetic retinopathy is theretinal microvasculature. The integrity of capillaryendothelial cells is crucial to maintain homeosta-

∗ Corresponding author. Tel.:+1-519-661-2030;fax: +1-519-661-2930.

E-mail address:[email protected] (S. Chakrabarti).

sis of the surrounding retinal tissue[1]. Endothelialcells produce and are responsive to the autocrine andparacrine activities of several vasoactive moleculeslike vascular endothelial growth factor (VEGF), en-dothelin (ET) and nitric oxide (NO)[2].

Diabetes increases the expression of VEGF sec-ondary to protein kinase C (PKC) activation[3]. TheVEGF protein family is comprised of several membersincluding VEGF A, B, C, D, E and placental growth

0168-8227/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.diabres.2004.02.002

198 M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208

factor [4]. VEGF mediates its activities through in-teractions with receptor tyrosine kinase proteins;VEGFR1 (Flt-1), VEGFR2 (KDR), or VEGFR3(Flt-4). A recently discovered co-receptor, neuropilinhas been demonstrated to associate with VEGFR2[5,6]. VEGF is an important factor promoting angio-genesis during proliferative diabetic retinopathy[7].However, since its discovery, VEGF has demonstratedthe ability to increase vascular tissue permeability innon-diabetic conditions[8].

Reduced microvascular blood flow and increasedvascular permeability are two early characteristic ab-normalities of diabetic microangiopathy[9]. Thereare reports outlining a possible pathogenetic role ofVEGF activation in early diabetes, where VEGF mayplay a role in the breakdown of the blood retinalbarrier [10,11]. Recently, it has been demonstratedthat VEGF neutralising antibody treatment is capableof preventing the diabetes induced increased per-meability [12]. VEGF also reacts extensively withother vasoactive factors. We have reported how anup-regulation of ET-1 and ET-3 mRNA levels inresponse to short term diabetes, contributed to re-duced blood flow in the retina[13]. We have furtherdemonstrated in human umbilical vein endothelialcells both ET-1 and VEGF may be responsible for theproduction of glucose induced, increased endothelialpermeability[14]. Furthermore, a co-stimulatory re-lationship between glucose induced ET and VEGFmay exist[15]. In addition, VEGF effector pathwaysinvolve increased nitric oxide synthase (NOS) mRNAexpression and NO production[16]. NO has beendemonstrated to cause a down-regulation of VEGF[17,18]. Interestingly, an up-regulation of ET canlead to a down-regulation of NO[19,20]. Hence,an intricate relationship may exist among thesefactors.

In order to delineate the pathogenetic mechanismsin early diabetic microangiopathy, the present studyinvestigated the role of VEGF and its interactionswith other vasoactive factors in the pathogenesis ofincreased microvascular permeability in the retinaof the streptozotocin (STZ) diabetic rat. We inves-tigated VEGF alterations in mediating retinal bloodflow changes in short term diabetes. Finally, we haveexamined the role of the MAP kinase signallingpathway in inducible nitric oxide synthase (iNOS)upregulation.

2. Materials and methods

2.1. Animals

All animals were cared for under the conditionsand rules designated by the ARVO Statement for theUse of Animals in Ophthalmic and Vision Researchwith approval by the University of Western OntarioAnimal Care and Ethics Committee. Male SpragueDawley rats of approximately 200 g received a singleintravenous injection of streptozotocin (65 mg/kg incitrate buffer, pH 6.5). Control animals received anequivalent injection of citrate buffer. After confir-mation of diabetes (blood glucose > 20 mmol/l on 2consecutive days), animals were randomised to oneof three treatment groups: poorly controlled diabet-ics, poorly controlled diabetics treated with high dose(30 mg/kg sub cutaneous daily) SU5416 (SUGEN Inc.San Francisco CA, USA courtesy of Drs. A. Howlett,N. Patel and G. McMahon) or poorly controlleddiabetics with low dose (20 mg/kg subcutaneous ev-ery second day) SU5416. Control animals receivedan equal volume injection of vehicle. SU5416 is aVEGFR2 signal inhibitor[21,22]. Animals were fol-lowed up for a 4 week treatment period with rat chowand water ad libitum. Animals were monitored forbody weight changes, glucosuria and ketouria. Alldiabetic rats were implanted with slow release insulinimplants to prevent ketosis (approximately 2 U/day)(LinShin, Scarborough, ON, Canada).

Prior to sacrifice, colour Doppler ultrasound analy-sis of the right eye was performed using previously de-scribed techniques[13]. Briefly, following anaesthesiawith ketamine and xylazine the central retinal vascula-ture was located with an 8L5, 80 MHz colour Dopplerultrasound probe (Acuson Mountainview, CA, USA).Doppler waveforms were examined and colour imageswere obtained in real time with Doppler spectral anal-ysis using the arterial tracings. At least three measure-ments were recorded for each animal and the mean wascalculated. The resistivity index (RI) of the central reti-nal artery was calculated by subtracting the diastolicvelocity (DV) from the peak systolic velocity (SV) anddividing by the systolic velocity[(SV−DV)/SV]. Anincreased central retinal artery RI value is indicative ofvasoconstriction at the capillary or pre-capillary levelwithin the retina[13]. Furthermore, colour Dopplerand echocardiographic studies on the heart were also

M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208 199

performed. RI values for the mitral valve and pul-monary artery were obtained. Tissue harvesting meantthe right retina was snap frozen in liquid nitrogenwhile the left retina was preserved in 10% bufferedformalin solution. Blood was collected for analysisof glucose levels (LifeScan, Burnaby, BC, Canada)glycated haemoglobin and plasma levels of SU5416.

2.2. RNA isolation

Using TRIZOLTM reagent (Invitrogen Inc. Burling-ton, ON, Canada), RNA was extracted with chloroform.Centrifugation separated the solution into aqueousand organic phases, where RNA was recovered fromthe aqueous phase by precipitation with isopropyl al-cohol. After suspension in DEPC-treated water, RNAwas quantified by measuring absorbance at 260 and280 nm. RNA samples were stored at−70◦C.

2.3. First strand cDNA synthesis

Using the Superscript-II system (Invitrogen Inc.)first strand cDNA was made. Five micrograms of RNAand oligo (dT) primers (Invitrogen Inc.) were dena-tured for 10 min at 65◦C, and the reaction was ter-minated by placing the samples on ice. Addition ofMMLV-reverse transcriptase and dNTP at 42◦C for50 min and a termination step of 15 min at 70◦C, pro-duced a final 20�l volume of RT product, which wasstored at−20◦C.

Table 1Rat primer sequences used for competitive RT-PCR

Primer name Competitor size (bp) Target size (bp) Sequence (5′→3′)

ET-1 484 500 (+) gct cct gct cct cct tga tggtac ggt cat cat ctg aca c(−) ctc gct cta tgt aag tca tgggcg tga gta tta cga cga agg tg

ET-3 294 383 (+) gca ctt gct tca ctt ata agggta cgg tca tca tct gac ac(−) aca gaa gca aga agc atc agt tgagag ttt ctg cgg cag tta a

eNOS 290 207 (+) gca aga ccg att aca cga cagtac ggt cat cat ctg aca c(−) gtc ctc agg agg tct tgc acagag ttt ctg cgg cag tta a

iNOS 301 220 (+) atg gaa cag tat aag cga aac accgta cgg tca tca tct gac ac(−) gtt tct ggt cga tgt cat gag caa aggaga gtt tct gcg gca gtt aa

VEGF 188/164/120 340 635/563/431 (+) ctg ctg tct tgg gtg cat tgggta cgg tca tca tct gac ac(−) cac cgc ctt ggc ttg tca catcgc cat cct ggg aag act cc

The complete sequence represents the primer sequence used to construct the competitor fragments, while the underlined sequence representsthe primers used to amplify the target gene.

2.4. Competitive RT-PCR

Due to the small amount of retinal tissues avail-able to perform multiple gene analysis, competitiveRT-PCR was used to quantify the expression of thespecific mRNAs. Competitive PCR amplifies a knownamount of sample (the competitor) to an unknownamount (the target) under the same reaction condi-tions. The method is superior to semi-quantitativePCR which is based on quantification of a housekeeping gene. Competitors were created using theTaKaRa DNA Construction Kit (Panvera, MadisonWI USA). Briefly, competitors were constructed usingthe primers found inTable 1. Sequences at both endsof the competitor were complementary to primersfor amplifying the target DNA. Competitors were ofcomparable size to target DNA fragments assuringsimilar amplification dynamics of both target andcompetitor DNA fragments. 20 pmol/l of sense andantisense primers along with 25�l of PCR water wereadded to the premix solution. Samples were amplifiedby 30 cycles consisting of 30 s steps at the temper-atures: 94, 60 and 72◦C. Samples were filtered withthe Suprec-02 cartridge and diluted with PCR waterto a final volume of 50�l. Copies per microliter werecalculated according to manufacturer’s instructions.Each competitor was diluted and preliminary experi-ments were performed to establish the optimal dilutionby measuring the product of the area and average in-tensity between competitor and target was a 1:1 ratio.

200 M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208

Amplification was performed with the primerslisted in Table 1. For all primer sets 1× PCR buffer,1.0�L RT product, 2.5�l of the appropriate dilutionof competitor, 500 nM of sense and antisense primers,0.25 mM of dNTP and 2.5 U of Platinum TaqTM (In-vitrogen Inc.) along with 2.0 mM MgCl2 for ET-1,ET-3, VEGF or 1.5 mM MgCl2 for endothelial nitricoxide synthase (eNOS), and iNOS to a final volume of25�l. 40 cycles of amplification were used for ET-3while all other genes were amplified for 30 cycles;92◦C for 45 s, 60◦C for 45 s and 72◦C for 1 min,followed by a final extension of 72◦C for 10 min.Preliminary experiments confirmed that amplificationwas in the linear phase of the PCR reaction.

2.5. Quantification

The PCR products were analysed on a 2% agarosegel in 1× TBE buffer. The gels were stained withethidium bromide and visualised with ultra violet light.For quantification, the ratio of optical density and areaof the target gene band and its competitor band wereassessed using MochaTM densitometry software (Jan-del Scientific, CA, USA). Due to competition of targetand competitor DNA to be amplified with the sameprimers, the ratio between the two amplified productsreflects the original amounts of target cDNA and itscompetitor.

2.6. Cell culture

To further explore a possible mechanism of iNOSexpression, endothelial cell culture experiments wereperformed with previously established techniqueswithin our laboratory.[23] A HUVEC cell line waspurchased from the American Type Tissue Collec-tion (Catalog # CRL-2480, Manassas, VA, USA).Cells were grown in 25 cm2 tissue culture flasks ina humidified 37◦C, 5% CO2 incubation chamber,with all experiments repeated in triplicate. VEGFsignal inhibitor SU5416 (Sugen Inc., 5 mM finalconcentration), MEK1/2 inhibitor U0126 (Promega,Madison WI, USA., 10�M final concentration), re-combinant human VEGF (Sigma, 50 ng ml−1 finalconcentration),d-glucose (25 mM final concentration)l-glucose (25 mM final concentration as a control)were added at 80% confluence. Cells were analysedafter 48 h of glucose incubation, as our previous stud-

ies have indicated that gene expression in endothelialcells occurs after such an incubation[14]. RNA wasextracted and analysed by real time PCR (see below).

2.7. Real time PCR

After RNA isolation and first strand cDNA prepara-tion, real time quantitative PCR using the Roche Light-Cycler system (Roche Diagnostics Canada, Lavalle,PQ, Canada) was performed. PCR reactions were car-ried out in mirco-capillary tubes (Roche DiagnosticsCanada) to a final volume of 20�l. The reaction mix-ture consisted of 1.0�l cDNA, 1.0�l both forwardand reverse 10�M primers, 1.6�l 25 mM MgCl2,5.4�l molecular grade H2O and 10.0�l SYBR GreenTaq ReadyMix, Capillary Formulation (Sigma AldrichCanada, Oakville ON, Canada).

The primer sequences for human�-actin were:sense 5′-CCTCTATGCCAACACAGTGC-3′, anti-sense 5′-CATCGTACTCCTGCTTGCTG-3′. The cy-cling parameters for�-actin were [temperature (◦C);hold time (s); ramp rate (◦C/s); denaturation step[95/30/20], followed by 35 cycles of amplification:denaturation [95/0/20], annealing [58/5/20], elonga-tion [72/8/20] and signal acquisition [83/1/20]. Melt-ing curve analysis consisted of a three step processwith continuous signal acquisition: step 1 [95/0/20],step 2 [67/15/20], step 3 [95/0/0.1]. The primer se-quences for human iNOS were: sense 5′-TGCAGAC-ACGTGCGTTACTCC-3′, anti-sense 5′-GGTAGCCA-GCATAGCGGATG-3′ [24]. The cycling parametersfor iNOS were: denaturation step [95/30/20], followedby 55 cycles of amplification: denaturation [95/0/20],annealing [57/5/20], elongation [72/9/20] with con-tinuous signal acquisition. The melting curve analysisconsisted of three steps with continuous signal acqui-sition: step 1 [95/0/20], step 2 [65/15/20] and step 3[95/0/0.1].

To optimise the amplification of the genes, melt-ing curve analysis was used to determine the meltingtemperature between specific products and primer–dimers. mRNA was quantified with the standard curvemethod. Standard curves for iNOS and�-actin wereconstructed using different amounts of standard tem-plate. The data was normalised to�-actin in order toaccount for differences in reverse transcription effi-ciencies and template amounts within each reactionmixture [25].

M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208 201

2.8. Immunohistochemistry

Retinal sections were immunocytochemically stai-ned for albumin. Five micrometer thick retinal tis-sue sections from formalin fixed, paraffin embeddedblocks were transferred to positively charged slides tobe used for staining. A polyclonal rabbit anti-humanalbumin antibody (DAKO Diagnostics. Mississauga,ON, Canada) (1:400) was used along with the strepta-vidin biotin reaction Vectastain Elite Kit (Vector Lab-oratories, Burlingame, CA, USA). Diaminobenzidinewas used as a chromogen. Slides were counterstainedwith hematoxylin. For negative controls, the primaryantibody was replaced with non-immune rabbit serumas well as PBS without immunoglobulin. The experi-ments were repeated three times with slides analysed,in a masked fashion by two investigators unaware ofthe treatment. Slides were arbitrarily scored as to theextravascular compartment staining using a gradedscale ranging from zero (no stain) to four (intensestain).

2.9. Statistical analysis

All values are displayed as mean± S.E.M. Valueswere analysed with ANOVA followed by analysis be-tween groups with the unpaired Student’st-test withBonferoni corrections. Qualitative data were analysedby the Chi-squared test. AP value of 0.05 or less wasconsidered significant.

3. Results

3.1. Clinical monitoring

All diabetic animals had higher blood glucoselevels, higher glycated haemoglobin levels and re-duced body weight gain compared to age matched,

Table 2Clinical parameters (average± S.E.M. for each group)

Non-diabetic control(n = 6)

Poorly controlleddiabetes (n = 6)

Poorly controlled diabetes withSU5416 (n = 12)

Body weight (g) 399.4± 10.66 347.9± 11.04a 288.3± 24.61a

Blood glucose (mmol/l) 5.5± 0.22 22.6± 1.15a 17.6 ± 3.92a

Glycated haemoglobin (%) 5.2± 0.77 12.2± 1.09a 9.8 ± 1.17a

a Denotes significant difference from non-diabetic control (P < 0.05).

non-diabetic control animals. Treatment with SU5416had no effect in modifying blood glucose and bodyweight changes [Table 2]. Twenty four hours postinjection plasma levels of SU5416 high dose groupwas 29.86 ± 2.75 ng ml−1, which is within the ther-apeutic range. SU5416 levels in the low dose groupwas determined to be 9.48± 2.75 ng ml−1.

3.2. Permeability alterations in the retina

In non-diabetic control animals albumin positivitywas limited to the lumen of the microvasculature,with very little or no positivity (range score 0–1)in the extravascular component. Poorly controlleddiabetic animals demonstrated increased immunore-activity throughout the retina and the intensity ofthe staining was greater in the extravascular region(range score 3–4) compared to control (P < 0.001).High dose SU5416 treatment prevented diabetes in-duced increased permeability and the immunostain-ing pattern was similar to the control animals (scorerange 0–1). No extravascular albumin staining wasseen in the lens or cornea, as an internal control(Fig. 1).

3.3. Blood flow alteration

Resistivity index as calculated from Dopplermeasurement, was significantly elevated in poorlycontrolled diabetic rats compared to non-diabeticanimals. High dose SU5416 therapy significantlyreduced the retinal RI value compared to poorly dia-betic animals and control animals. Low dose SU5416treatment however failed to achieve a therapeutic ef-fect. (Fig. 2) No significant changes in either mitralvalve or pulmonary artery RI values due to dia-betes or SU5416 were observed. [RI of pulmonaryartery was 0.88 ± 0.01 (control), 0.89 ± 0.02 (dia-betes), 0.94±0.02 (diabetes with high dose SU5416),

202 M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208

Fig. 1. Immuno-histochemical staining of the retina for albumin to detect capillary permeability. (a) From a non-diabetic control rat, (b)poorly controlled diabetic rat, (c) poorly controlled diabetic rat with SU5416 treatment and (d) negative control. No permeability alterationwas seen in the adjoining lens (L). (Arrows indicate microvessels, original magnification of (a); (b); (d)= 100×; (c) = 200× (n = 6 forsamples a, b, c.)

0.91±0.01 (diabetes with low dose SU5416). RI of themitral valve was 0.91±0.01 (control), 0.94±0.01 (di-abetes), 0.94±0.00 (diabetes with high dose SU5416),0.91± 0.03 (diabetes with low dose SU5416).]

3.4. Retinal mRNA expression

In vivo, poorly controlled diabetes increased themRNA expression of ET-1, ET-3, and VEGF 164 inthe retina. eNOS, iNOS, VEGF 120 and VEGF 188expression remained unchanged in poorly controlleddiabetes. SU5416 treatment had no effect on ET-3 andeNOS mRNA expression in diabetic animals. How-ever, iNOS expression was significantly augmented by

both doses of SU5416 treatment. SU5416 high dosefurther augmented VEGF 120 and VEGF 164 mRNAexpression in the retina (Fig. 3). The ET-1 mRNAexpression following treatment of diabetic animalswith both low and high doses of SU5416 remainedsignificantly elevated compared to non-diabetic ani-mals. Although the levels were slightly higher thanuntreated diabetic animals, statistical significance wasnot achieved.

3.5. Mechanism of SU5416 iNOS upregulation

As our studies on the retina demonstrated an in-ducible effect of SU5416 on iNOS expression (Fig. 3),

M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208 203

Fig. 2. Resistivity index (RI) as calculated from colour Doppler ultrasound of the central retinal artery (seeSection 2). Poorly controlleddiabetes significantly increased the RI value, while treatment of diabetic rats with high dose SU5416 significantly reduced the RI. Lowdose SU5416 had no effect on RI. (n = 6 for all groups.) (∗) Indicates significant difference from non-diabetic controls (P < 0.05); (†)indicates significant difference from poorly controlled diabetes (P < 0.05).

we further explored the mechanism of such phe-nomenon using cultured endothelial cells. Previousstudies in kidney and neuronal tissues[26–28]showediNOS mRNA expression may be partially mediatedthrough a MAP kinase pathway. To determine whetheriNOS mRNA upregulation due to SU5416 occurredthrough a MAPK signalling pathway, selective inhi-bition of MAPK was used in a cell culture (HUVEC)system. HUVEC cells treated with 25 mM glucosesuppressed iNOS expression. However, SU5416 inaddition to 25 mM glucose did increase iNOS ex-pression. The MAPK inhibitor U0126 suppressed theexpression of iNOS in spite of SU5416 treatment(Fig. 4).

4. Discussion

In the present study we demonstrated the impor-tant role of VEGF by itself and its interaction withother vasoactive substances in the pathogenesis of di-abetes induced early functional changes in the retina.Increased vascular permeability and microvascularblood flow abnormalities are characteristic featuresof diabetic retinopathy[1,2]. VEGF is a potent factorcausing increased vascular permeability. Increasesin vascular permeability have been demonstratedin STZ diabetic rats after 1 week of diabetes[29].Quantitatively, increased vascular permeability asdemonstrated by increased albumin permeation hasbeen shown to increase 2.9 fold after 1 week and 10.7

fold after 4 weeks of diabetes[30]. In this study wehave demonstrated that increased permeability occursin association with increased VEGF mRNA expres-sion in the retina and such increases in permeabilitymay be prevented using a specific blocker of VEGFsignalling. SU5416 was developed as a chemothera-peutic agent used to inhibit angiogenesis[21,22,31].This compound has demonstrated to be very effec-tive in blocking VEGF signalling in other systems[21,22,30,31]. Furthermore, SU5416 has demon-strated an ability to cross the blood brain barrier[32].However, this is the first report where this compoundhas been used to prevent diabetes induced microvas-cular abnormalities. The method used for demon-stration of increased permeability is simple, reliableand has been used previously by other investigators[33].

VEGF may interact with ICAM-1 in mediatingincreased permeability in diabetes[30]. ICAM-1 ex-pression can be induced through VEGF stimulation[34,35]. Increased ICAM-1 expression is associatedwith a breakdown of the blood retinal barrier by wayof increased leukocyte adhesion[30]. Furthermore,the inhibition of ICAM-1 has demonstrated a re-duction of the blood retina barrier breakdown[30].PKC activation is an important mechanism leadingto VEGF up-regulation in diabetes[7,11,36]. How-ever, it has previously been shown that augmentedpolyol pathway activity, non enzymatic glycation andoxidative stress may also regulate VEGF expression[37].

204 M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208

Fig. 3. Diagrammatic representation of retinal mRNA expression of (a) ET-1; (b) ET-3; (c) iNOS; (d) eNOS and (e) VEGF. Diabetes leadto increased expression of ET-1, ET-3 and VEGF mRNA but not eNOS or iNOS mRNA. SU5416 treatment on the other hand lead toincreased iNOS mRNA expression. (n = 6 for all groups.) (∗) Indicates significant difference from non-diabetic controls (P < 0.05).

M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208 205

Fig. 4. Diagrammatic representation of iNOS mRNA analysis from HUVEC cells as analysed by real time PCR. (a) Treatment of cellswith 25 mM d-glucose lead to decreased iNOS mRNA expression. SU5416 treatment along with 25 mMd-glucose lead to increased iNOSmRNA expression which was normalised by treatment with a MAPK inhibitor. (b) Represents an amplification plot of human iNOS. (c)Represents a melting curve analysis plot from human iNOS demonstrating the specificity of the PCR product. (∗) Indicates significantdifference from 5 mM glucose (P < 0.05); (†) indicates significant difference from 25 mM glucose treatment (P < 0.05).

206 M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208

It is of further interest that this study has demon-strated an interaction of VEGF with other vasoactivefactors which are important in mediating blood flowwithin the retina. We and others have previouslydemonstrated increased ET-1 expression which is akey factor in the production of retinal vasoconstric-tion [13,38,39]. We used colour Doppler ultrasoundbased RI measurement as an indicator of retinalcapillary vasoconstriction. As RI values across themitral valve or pulmonary artery were not altered, thepresent data suggest that changes seen in this studyreflect retinal microvasculature changes. Diabetic an-imals as previously reported, demonstrated increasedRI in the retina, in association with increased ETexpression [13,38]. VEGF signal inhibition withhigh dose SU5416 was found to abolish diabetes in-duced retinal vasoconstriction in spite of increasedET-1 expression. Interestingly, the same treatmentalso lead to increased iNOS expression. Hence, it isconceptually possible that increased NO productionsecondary to SU5416 treatment may have counter-balanced the ET induced vasoconstriction. On theother hand, animals treated with low dose SU5416did not normalise the diabetes-induced, increasedRI. Although iNOS mRNA expression of high andlow dose SU5416 were similar, the ET-1 mRNAlevel was highest in the low dose SU5416 animals.The exact cause of this phenomenon is not clear.Due to the intricate regulatory mechanisms betweenthese vasoactive factors, it is theoretically possiblethat partial VEGF signal blockade with SU5416 mayhave further augmented ET-1 expression which pre-vented the vasodilatory effects of NO. However, it isalso possible that other yet unidentified mechanismswhich are not mediated by NO may also be responsi-ble. These ideas will require further confirmation. Ithas been demonstrated that VEGF may increase NOproduction[16,17]. We have previously ascertainedthat although treatment of non-diabetic animals withNO blocker produced no change in RI, the treat-ment of diabetic animals with NO donor preventeda diabetes induced RI increase[40]. In addition, ETand VEGF have a co-stimulatory relationship[15].In the present study however, VEGF signal inhibi-tion was not effective in preventing diabetes inducedET-1 up-regulation in the retina, suggesting severalmechanisms may be involved in the up-regulationof ETs in diabetes[41]. As eluded to earlier ET-1

and NO also have a counter-inhibitory relationship[19,42].

To further characterise the mechanism of increasediNOS mRNA expression observed in the diabetic ani-mals treated with SU5416 we investigated endothelialcells with respect to MAPK signalling. Mitogen acti-vated protein kinase (MAPK) signalling consists of athree protein system signal cascade: the MAPK proteinand upstream MAPK kinase (MAPKK) and MAPKKkinase (MAPKKK)[43]. There are three MAPK path-ways: the extracellular signal regulated kinases (erk1and 2), the c-Jun-N terminal kinase 1 (JNK 1) and thep38 MAPK [44]. In diabetes, MAPK proteins are ac-tivated due to the direct effects of hyperglycemia, glu-cose induced oxidative stress, and AGE interactions[45–48]. In our endothelial cell system, the SU5416induced up-regulation of iNOS mRNA was normalisedwith the use of a MAPK inhibitor, suggesting that theeffects of VEGF receptor signal inhibition may be par-tially mediated through a MAPK signal cascade. Inkeeping with this notion it has previously been demon-strated that MAPK signalling is important in the reg-ulation of NOS expression[26].

In summary, we have demonstrated that VEGF isan important mediator of increased vascular perme-ability in diabetes. In addition, VEGF by its interac-tion with ET and NO may be of importance in me-diating retinal blood flow abnormalities in diabetes.Furthermore, MAPK signalling may be significant inthe regulation of iNOS. VEGF signal inhibition mayoffer an important therapeutic modality for diabeticretinopathy.

Acknowledgements

This work was supported in part by grants from theCanadian Diabetes Association in memory of GlennW. Liebrock as well as the Lawson Health ResearchInstitute Internal Research Fund. The authors wish tothank K. Mukherjee for the histological preparationsand immunostaining.

References

[1] G.L. King, M. Brownlee, The cellular and molecular mecha-nisms of diabetic complications, Endocrinol. Metab. Clin.North Am. 25 (1996) 255–270.

M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208 207

[2] S. Chakrabarti, M. Cukiernik, D. Hileeto, T. Evans, S. Chen,Role of vasoactive factors in the pathogenesis of early changesin diabetic retinopathy, Diabetes Metab. Res. Rev. 16 (2000)393–407.

[3] I. Spyridopoulos, C. Luedemann, D. Chen, M. Kearney,D. Chen, T. Murohara, N. Principe, J.M. Isner, D.W.Losordo, Divergence of angiogenic and vascular permeabilitysignalling by VEGF: inhibition of protein kinase C suppressesVEGF-induced angiogenesis, but promotes VEGF-induced,NO-dependent vascular permeability, Arterioscler. Thromb.Vasc. Biol. 22 (2002) 901–906.

[4] T.V. Petrova, T. Makinen, K. Alitalo, Signalling via vascularendothelial growth factor receptors, Exp. Cell Res. 253 (1999)117–130.

[5] L.P. Aiello, J.S. Wong, Role of vascular endothelial growthfactor in diabetic vascular complications, Kidney Int. Suppl.77 (2000) S113–S119.

[6] N. Ferrara, Role of vascular endothelial growth factor inregulation of physiological angiogenesis, Am. J. Physiol. CellPhysiol. 280 (2001) C1358–C1366.

[7] L.P. Aiello, S.E. Bursell, A. Clermont, E. Duh, H. Ishii, C.Takagi, F. Mori, T.A. Ciulla, K. Ways, M. Jirousek, L.E.Smith, G.L. King, Vascular endothelial growth factor-inducedretinal permeability is mediated by protein kinase C in vivoand suppressed by an orally effective beta-isoform-selectiveinhibitor, Diabetes 46 (1997) 1473–1480.

[8] D.R. Senger, S.J. Galli, A.M. Dvorak, C.A. Perruzzi,V.S. Harvey, H.F. Dvorak, Tumor cells secrete a vascularpermeability factor that promotes accumulation of ascitesfluid, Science 219 (1983) 983–985.

[9] C.A. Do, P. Ramos, A. Reis, R. Proenca, J.G. Cunha-vaz,Breakdown of the inner and outer blood retinal barrier instreptozotocin-induced diabetes, Exp. Eye Res. 67 (1998)569–575.

[10] T. Murata, K. Nakagawa, A. Khalil, T. Ishibashi, H. Inomata,K. Sueishi, The relation between expression of vascularendothelial growth factor and breakdown of the blood-retinalbarrier in diabetic rat retinas, Lab. Invest. 74 (1996) 819–825.

[11] P. Xia, L.P. Aiello, H. Ishii, Z.Y. Jiang, D.J. Park, G.S.Robinson, H. Takagi, W.P. Newsome, M.R. Jirousek, G.L.King, Characterization of vascular endothelial growth factor’seffect on the activation of protein kinase C, its isoforms,and endothelial cell growth, J. Clin. Invest. 98 (1996) 2018–2026.

[12] R.G. Tilton, T. Kawamura, K.C. Chang, Y. Ido, R.J.Bjercke, C.C. Stephan, T.A. Brock, J.R. Williamson, Vasculardysfunction induced by elevated glucose levels in rats ismediated by vascular endothelial growth factor, J. Clin. Invest.99 (1997) 2192–2202.

[13] D. Deng, T. Evans, K. Mukherjee, D. Downey, S. Chakrabarti,Diabetes-induced vascular dysfunction in the retina: role ofendothelins, Diabetologia 42 (1999) 1228–1234.

[14] S. Chen, M.D. Apostolova, M.G. Cherian, S. Chakrabarti,Interaction of endothelin-1 with vasoactive factors in media-ting glucose-induced increased permeability in endothelialcells, Lab. Invest. 80 (2000) 1311–1321.

[15] A. Cruz, C. Parnot, D. Ribatti, P. Corvol, J.M. Gasc,Endothelin-1, a regulator of angiogenesis in the chick chori-oallantoic membrane, J. Vasc. Res. 38 (2001) 536–545.

[16] J. Kroll, J. Waltenberger, VEGF-A induces expression ofeNOS and iNOS in endothelial cells via VEGF receptor-2(KDR), Biochem. Biophys. Res. Commun. 252 (1998) 743–746.

[17] A. Bouloumie, V.B. Schini-Kerth, R. Busse, Vascularendothelial growth factor up-regulates nitric oxide synthaseexpression in endothelial cells, Cardiovasc. Res. 41 (1999)773–780.

[18] N. Ghiso, R.M. Rohan, S. Amano, R. Garland, A.P. Adamis,Suppression of hypoxia-associated vascular endothelialgrowth factor gene expression by nitric oxide via cGMP,Invest. Ophthalmol. Vis. Sci. 40 (1999) 1033–1039.

[19] E.R. Levin, Endothelins, N. Engl. J. Med. 333 (1995) 356–363.

[20] P.M. Vanhoutte, Endothelin-1. A matter of life and breath,Nature 368 (1994) 693–694.

[21] T.A. Fong, L.K. Shawver, L. Sun, C. Tang, H. App, T.J.Powell, Y.H. Kim, R. Schreck, X. Wang, W. Risau, A. Ullrich,K.P. Hirth, G. McMahon, SU5416 is a potent and selectiveinhibitor of the vascular endothelial growth factor receptor(Flk-1/KDR) that inhibits tyrosine kinase catalysis, tumorvascularization, and growth of multiple tumor types, CancerRes. 59 (1999) 99–106.

[22] L. Angelov, B. Salhia, L. Roncari, G. McMahon, A.Guha, Inhibition of angiogenesis by blocking activation ofthe vascular endothelial growth factor receptor 2 leads todecreased growth of neurogenic sarcomas, Cancer Res. 59(1999) 5536–5541.

[23] S. Chen, S. Mukherjee, C. Chakraborty, S. Chakrabarti, Highglucose-induced, endothelin-dependent fibronectin synthesisis mediated via NF-kappa B and AP-1, Am. J. Physiol. CellPhysiol. 284 (2003) C263–C272.

[24] J. Dotsch, N. Hogen, Z. Nyul, J. Hanze, I. Knerr, M.Kirschbaum, W. Rascher, Increase of endothelial nitric oxidesynthase and endothelin-1 mRNA expression in humanplacenta during gestation, Eur. J. Obstet. Gynecol. Reprod.Biol. 97 (2001) 163–167.

[25] S. Chen, Z.A. Khan, M. Cukiernik, S. Chakrabarti, Diffe-rential activation of NF-kappa B and AP-1 in increasedfibronectin synthesis in target organs of diabetic compli-cations, Am. J. Physiol. Endocrinol. Metab. 284 (2003)E1089–E1097.

[26] M. Poljakovic, J.M. Nygren, K. Persson, Signalling pathwaysregulating inducible nitric oxide synthase expression in humankidney epithelial cells, Eur. J. Pharmacol. 469 (2003) 21–28.

[27] A. Lahti, H. Kankaanranta, E. Moilanen, P38 mitogen-acti-vated protein kinase inhibitor SB203580 has a bi-directionaleffect on iNOS expression and NO production, Eur. J. Phar-macol. 454 (2002) 115–123.

[28] I.O. Han, H.S. Kim, H.C. Kim, E.H. Joe, W.K. Kim,Synergistic expression of inducible nitric oxide synthase byphorbol ester and interferon-gamma is mediated through NF-kappaB and ERK in microglial cells, J. Neurosci. Res. 73(2003) 659–669.

208 M. Cukiernik et al. / Diabetes Research and Clinical Practice 65 (2004) 197–208

[29] T. Qaum, Q. Xu, A.M. Joussen, M.W. Clemens, W. Qin,K. Miyamoto, H. Hassessian, S.J. Wiegand, J. Rudge,G.D. Yancopoulos, A.P. Adamis, VEGF-initiated blood-retinalbarrier breakdown in early diabetes, Invest. Ophthalmol. Vis.Sci. 42 (2001) 2408–2413.

[30] K. Miyamoto, S. Khosrof, S.E. Bursell, R. Rohan, T. Murata,A.C. Clermont, L.P. Aiello, Y. Ogura, A.P. Adamis, Preventionof leukostasis and vascular leakage in streptozotocin-induceddiabetic retinopathy via intercellular adhesion molecule-1inhibition, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 10836–10841.

[31] D.B. Mendel, R.E. Schreck, D.C. West, G. Li, L.M. Strawn,S.S. Tanciongco, S. Vasile, L.K. Shawver, J.M. Cherrington,The angiogenesis inhibitor SU5416 has long-lasting effects onvascular endothelial growth factor receptor phosphorylationand function, Clin. Cancer Res. 6 (2000) 4848–4858.

[32] T. Takamoto, M. Sasaki, T. Kuno, N. Tamaki, Flk-1 specifickinase inhibitor (SU5416) inhibited the growth of GS-9Lglioma in rat brain and prolonged the survival, Kobe J. Med.Sci. 47 (2001) 181–191.

[33] S.A. Vinores, C. Gadegbeku, P.A. Campochiaro, W.R. Green,Immunohistochemical localization of blood-retinal barrierbreakdown in human diabetics, Am. J. Pathol. 134 (1989)231–235.

[34] M. Lu, V.L. Perez, N. Ma, K. Miyamoto, H.B. Peng, J.K.Liao, A.P. Adamis, VEGF increases retinal vascular ICAM-1expression in vivo, Invest. Ophthalmol. Vis. Sci. 40 (1999)1808–1812.

[35] K. Miyamoto, S. Khosrof, S.E. Bursell, Y. Moromizato,L.P. Aiello, Y. Ogura, A.P. Adamis, Vascular endothelialgrowth factor (VEGF)-induced retinal vascular permeabilityis mediated by intercellular adhesion molecule-1 (ICAM-1),Am. J. Pathol. 156 (2000) 1733–1739.

[36] P. Xia, T. Inoguchi, T.S. Kern, R.L. Engerman, P.J. Oates,G.L. King, Characterization of the mechanism for the chronicactivation of diacylglycerol–protein kinase C pathway indiabetes and hypergalactosemia, Diabetes 43 (1994) 1122–1129.

[37] I.G. Obrosova, A.G. Minchenko, V. Marinescu, L. Fathallah,A. Kennedy, C.M. Stockert, R.N. Frank, M.J. Stevens,Antioxidants attenuate early up regulation of retinal vascularendothelial growth factor in streptozotocin-diabetic rats,Diabetologia 44 (2001) 1102–1110.

[38] T. Evans, D.D. Xi, K. Mukherjee, D. Downey, S. Chakrabarti,Endothelins, their receptors, and retinal vascular dysfunctionin galactose-fed rats, Diabetes Res. Clin. Pract. 48 (2000)75–85.

[39] S. Dallinger, G.T. Dorner, R. Wenzel, U. Graselli, O.Findl, H.G. Eichler, M. Wolzt, L. Schmetterer, Endothelin-1contributes to hyperoxia-induced vasoconstriction in thehuman retina, Invest. Ophthalmol. Vis. Sci. 41 (2000) 864–869.

[40] M. Cukiernik, S. Mukherjee, D. Downey, S. Chrakabarti,Heme oxygenase in the retina in diabetes, Curr. Eye Res. 27(2003) 301–308.

[41] S. Chakrabarti, M. Cukiernik, S. Mukherjee, S. Chen,Therapeutic potential of endothelin receptor antagonists indiabetes, Expert. Opin. Investig. Drugs 9 (2000) 2873–2888.

[42] D. Fukumura, T. Gohongi, A. Kadambi, Y. Izumi, J. Ang,C.O. Yun, D.G. Buerk, P.L. Huang, R.K. Jain, Predominantrole of endothelial nitric oxide synthase in vascular endothelialgrowth factor-induced angiogenesis and vascular permeability,Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 2604–2609.

[43] S. Kumar, J. Boehm, J.C. Lee, p38 MAP kinases: keysignalling molecules as therapeutic targets for inflammatorydiseases, Nat. Rev. Drug Discov. 2 (2003) 717–726.

[44] G. Pearson, F. Robinson, G.T. Beers, B.E. Xu, M. Karandikar,K. Berman, M.H. Cobb, Mitogen-activated protein (MAP)kinase pathways: regulation and physiological functions,Endocr. Rev. 22 (2001) 153–183.

[45] D.M. Cohen, Mitogen-activated protein kinase cascades andthe signalling of hyperosmotic stress to immediate earlygenes, Comp. Biochem. Physiol. A Physiol. 117 (1997) 291–299.

[46] D.R. Tomlinson, Mitogen-activated protein kinases as glucosetransducers for diabetic complications, Diabetologia 42 (1999)1271–1281.

[47] M. Haneda, S. Araki, M. Togawa, T. Sugimoto, M. Isono, R.Kikkawa, Mitogen-activated protein kinase cascade is acti-vated in glomeruli of diabetic rats and glomerular mesangialcells cultured under high glucose conditions, Diabetes 46(1997) 847–853.

[48] X. Wang, J.L. Martindale, Y. Liu, N.J. Holbrook, The cellularresponse to oxidative stress: influences of mitogen-activatedprotein kinase signalling pathways on cell survival, Biochem.J. 333 (Pt 2) (1998) 291–300.