methylglyoxalinduced imbalance in the ratio of vascular … · exp physiol 00.00 pp 1–16 1...

Exp Physiol 00.00 pp 1–16 1

Experimental Physiology – Research Paper

Methylglyoxal-induced imbalance in the ratio of vascularendothelial growth factor to angiopoietin 2 secreted byretinal pigment epithelial cells leads to endothelialdysfunction

C. F. Bento1,2, R. Fernandes1,3, P. Matafome1,4, C. Sena4, R. Seica4 and P. Pereira1

1Center of Ophthalmology and Vision Sciences (COCV) - IBILI, Faculty of Medicine, University of Coimbra, Coimbra, Portugal2BEB PhD Programme, Centre for Neuroscience and Cell Biology (CNC), University of Coimbra, Coimbra, Portugal3Institute of Pharmacology and Experimental Therapeutics - IBILI, Faculty of Medicine, University of Coimbra, Coimbra, Portugal4Institute of Physiology - IBILI, Faculty of Medicine, University of Coimbra, Coimbra, Portugal

Progressive microvascular complications are a main feature of diabetes and are associatedwith impairment of the angiogenic response. Methylglyoxal (MGO) has been implicated in themolecular events that lead to endothelial dysfunction in diabetes. In this study, we hypothesizethat increased levels of MGO disrupt the ratio of vascular endothelial growth factor (VEGF)to angiopoietin 2 (Ang 2) secreted by retinal pigment epithelial (RPE) cells, which provides akey destabilizing signal that leads to apoptosis and decreased proliferation of retinal endothelialcells. Indeed, we show that MGO increases the levels of Ang 2 and dramatically decreases thelevels of VEGF secreted by RPE cells in response to hypoxia. Downregulation of VEGF islikely to be related to decreased hypoxia-inducible factor-1α (HIF-1α) protein levels and HIF-1transcriptional activity. Data further show that MGO-induced imbalance in the VEGF/Ang IIratio significantly changes the levels of BAX and Bcl-2 in endothelial cells. Moreover, thisimbalance is accompanied by an increase in the activity of caspase-3 and decreased proliferationof endothelial cells. Data obtained in cell culture systems are consistent with observations inretinas of diabetic animals, where increased availability of MGO is associated with changes indistribution and levels of HIF-1α, VEGF and Ang 2 and increased microvascular permeability. Inconclusion, the MGO-induced imbalance in the VEGF/Ang 2 ratio secreted by retinal epithelialcells activates apoptosis and decreases proliferation of retinal endothelial cells, which are likelyto contribute to endothelial dysfunction in diabetic retinopathy.

(Received 13 April 2010; accepted after revision 18 June 2010; first published online 18 June 2010)Corresponding author P. Pereira: COCV – IBILI, Azinhaga de Santa Comba, Celas, 3004-548 Coimbra, Portugal.Email: [email protected]

Vascular dysfunction is a main feature of diabetes andinvolves both micro- and macroangiopathy (Cooper et al.1997). Progressive and long-term vascular complicationsof diabetes include cardiovascular disease, chronic renalfailure, retinal damage, neuropathy and poor woundhealing. The majority of these vascular complications areassociated, in a more or less obvious way, with an impairedangiogenic response (Shoji et al. 2006). For example,although late stages of diabetic retinopathy (DR) areassociated with neovascularization, the disease originateswith endothelial cell damage, extensive capillary regressionand widespread inner retinal ischaemia (Engerman, 1989;

Cai & Boulton, 2002; Hammes et al. 2002). Indeed,studies in retinal capillaries have revealed that the lossof pericytes and formation of acellular capillaries are earlymorphological changes in the diabetic retina (Engerman,1989; Cai & Boulton, 2002; Hammes et al. 2002); however,the molecular mechanisms that underlie endothelialdysfunction in diabetes are still unclear.

Among the plethora of hyperglycaemia-mediatedmechanisms proposed to induce vascular dysfunction indiabetes, there is increasing evidence that accumulationof methylglyoxal (MGO) and formation of advancedglycation end-products (AGEs) may contribute to

C© 2010 The Authors. Journal compilation C© 2010 The Physiological Society DOI: 10.1113/expphysiol.2010.053561

2 C. F. Bento and others Exp Physiol 00.00 pp 1–16

the impaired angiogenic response (Stitt et al. 2005).Methylglyoxal is a highly reactiveα-oxoaldehyde formed asa byproduct of glycolysis, and its production is enhancedby exposure of cells to high levels of glucose (Kalapos, 1999;Ramasamy et al. 2006). Methylglyoxal is associated withthe formation of acellular capillaries and pericyte drop-out in diabetic rats (Hammes, 2003; Hammes et al. 2003).Moreover, it has been shown that AGEs attenuate theangiogenic response in vitro (Teixeira & Andrade, 1999),while in diabetic mice in vivo, inhibition of the formationof AGEs can restore ischaemia-induced angiogenesis inperipheral limbs (Tamarat et al. 2003). Interestingly,MGO increases the expression of angiopoietin 2 (Ang 2;Hammes et al. 2004; Yao et al. 2007) in diabetic retinas,which in the absence of pro-angiogenic factors inducesvessel destabilization and regression, as observed inmodels of solid tumours (Maisonpierre et al. 1997; Holashet al. 1999; Yancopoulos et al. 2000). The function of Ang 2appears to be highly dependent on the presence of vascularendothelial growth factor (VEGF; Maisonpierre et al. 1997;Holash et al. 1999; Oshima et al. 2004; Feng et al. 2008).For example, increased levels of Ang 2 in the presenceof low concentrations of VEGF were shown to result inendothelial cell death and vessel regression, whereas inthe presence of high levels of VEGF there is increasedproliferation and stimulation of angiogenesis.

In this study, we show that exposure of retinal pigmentepithelial (RPE) cells to MGO leads to destabilizationof hypoxia-inducible factor-1α (HIF-1α), which is a keyregulator of VEGF. Consistently, treatment with MGOresulted in decreased transcriptional activity of HIF-1and decreased levels of VEGF secreted by epithelial cells.Conversely, data also showed that MGO led to an increasein the levels of Ang 2 secreted by these cells. The netresult was a dramatic decrease in the ratio between VEGFand Ang 2 secreted into the culture medium. This, inturn, led to a significant decrease in the proliferation ofendothelial cells incubated in preconditioned media, aswell as increased activity of caspase-3. Data obtained incell culture systems was consistent with observations inretinas of a diabetic animal model. Altogether, the datasuggest that the VEGF/Ang 2 ratio imbalance inducedby MGO activates the apoptotic cascade and decreasesproliferation of endothelial cells, possibly contributing tovascular dysfunction in situations of increased availabilityof MGO, such as diabetes.

Methods

Cell culture and treatments

The retinal pigment epithelial cell line ARPE-19 (LGCPromochem, Teddington, UK) was cultured in Ham’sF12/Dulbecco’s modified Eagle’s medium (DMEM; 1:1)supplemented with 10% fetal bovine serum, antibiotics

(100 U ml−1 penicillin and 100 μg ml−1 streptomycin)and GlutaMax (1×). This cell line was maintained at37◦C in an atmosphere of air containing 5% CO2. Theconditionally immortalized rat retinal endothelial cell lineTR-iBRB (Hosoya et al. 2001) was a kind gift from Dr K.Hosoya (Graduate School of Medicine and PharmaceuticalSciences, University of Toyama, Japan). The TR-iBRBcells were maintained at 33◦C in an atmosphere ofair containing 5% CO2 in DMEM with low glucose(5.5 mM), supplemented with 10% fetal bovine serum,100 U ml−1 penicillin and 100 μg ml−1 streptomycin. Allmedia, GlutaMax and non-essential amino acids werepurchased from Invitrogen (Carlsbad, CA, USA). Whenappropriate, cells were treated as indicated with thefollowing agents: methylglyoxal (MGO; Sigma-Aldrich, StLouis, MO, USA), recombinant VEGF165 and recombinantAng 2 (R&D Systems, Minneapolis, MN, USA). For thehypoxic treatments (2% O2), a Nuaire N4950E incubator(Nuaire, Plymouth, MN, USA) was used.

Preconditioning protocol

Subconfluent ARPE-19 cells were treated as describedabove. After the treatments, cell culture media werecollected and centrifuged for 10 min at 5000g (to pelletcell debris). Subsequently, MGO that might remain inthe media was eliminated by filtration using centricons(with a cut-off of 10 kDa; Millipore, Bedford, MA, USA).For this purpose, the components of the preconditionedmedia (3 ml total volume) were concentrated to a volumeof 100 μl and subsequently resuspended to a final volumeof 3 ml of fresh medium. Levels of MGO in the mediawere assessed by HPLC. Levels of MGO that remainedin the media after centrifugation and resuspension wereundetectable. The preconditioned media were used toincubate TR-iBRB cells for 24 or 48 h. At the beginningof the incubation, TR-iBRB cells were at 50% (for24 h of incubation) or 30% of confluence (for 48 h ofincubation).

Animal models

Male diabetic Goto–Kakizaki (GK) and control Wistarrats (6 months old) were obtained from local breedingcolonies maintained at the animal facility of the Facultyof Medicine, University of Coimbra. The animals weresubjected to a constant daily cycle of 12 h light and 12 hdark with constant temperature (22–24◦C) and humidity(50–60%). Rats were given free access to water and tostandard commercial pelleted chow diet (AO4; Panlab,Barcelona, Spain). The animals were anaesthetized usingketamine chloride (75 mg kg−1, I.M., Parke-Davis, AnnArbor, MI, USA) and chlorpromazine chloride (2.65 mgkg−1, I.M., LabVitoria, Amadora, Portugal) and were killedby cervical dislocation. The eyes were enucleated and

C© 2010 The Authors. Journal compilation C© 2010 The Physiological Society

Exp Physiol 00.00 pp 1–16 Methylglyoxal imbalances the VEGF/Angiopoietin 2 ratio 3

retinas isolated. Retinas were frozen in liquid nitrogenor embedded in OCT tissue embedding matrix (ThermoScientific, Waltham, MA, USA) at −45◦C for subsequentcryosectioning. Retinas used for HPLC or Western blottingwere homogenized with a Potter Elvehjem homogenizer in300 μl of lysis buffer (50 mM Tris–HCl (pH 7.4), 250 mM

NaCl and 1× protease inhibitor cocktail), incubated for1 h on ice and briefly sonicated. In the case of lysatesused for HPLC, protein concentration was determinedand normalized. Subsequently, 100 μl of sample (at aconcentration of 1 μg μl−1) was mixed with 100 μl HCl(0.2 M), and the subsequent procedures were performedas explained in the online Supplemental Methods. In thecase of samples used for Western blotting, cell lysateswere centrifuged at 16,000g for 10 min at 4◦C, and thesupernatants were transferred to new tubes and were usedto determine protein concentration. The concentration ofall samples was normalized, and samples were denaturatedwith 2× Laemmli buffer, boiled at 100◦C for 5 min andused for immunoblotting, as explained in the onlineSupplemental Methods. All experiments involving animalswere performed in accordance with the ARVO Statementfor the Use of Animals in Ophthalmic and VisionResearch.

Antibodies and plasmids

The following antibodies were used in this work: mouseanti-HIF-1α clone H1alpha67 (Abcam, Cambridge,UK), mouse anti-actin clone C4 (Millipore-Chemicon,Billerica, MA, USA), mouse anti-Bcl-2 clone C-2 (SantaCruz Biotechnology, Santa Cruz, CA, USA), mouse anti-V5 tag clone 2F11F7 (Invitrogen), rat anti-GLO1 clone6F10 (Abcam), rabbit anti-VEGF (Ab-4; Calbiochem, SanDiego, CA, USA), rabbit anti-Ang 2 (Abcam), rabbit anti-BAX (�21; Santa Cruz Biotechnology), rabbit anti-Ki67(SP6; Abcam) and rabbit anti-von Willebrand Factor(vWF; Dako, Glostrup, Denmark). The plasmids pAD/V5-hGLOI and pcDNA6.2-GW/EmGFP-mIR hGLOI werekindly provided by Drs C. Marques (IBILI – Facultyof Medicine, University of Coimbra, Portugal) andJ. Ramalho (CEDOC – Faculty of Medical Sciences,New University of Lisbon, Portugal). The pT81 HRE-luciferase plasmid was kindly provided by Dr S. B. Catrina(Karolinska Institute, Stockholm, Sweden).

Luciferase assay

ARPE-19 cells were plated in 24-well plates atsubconfluent density and transfected with pT81 HRE-luciferase (Catrina et al. 2004), using Lipofectamine 2000(Invitrogen). The plasmid pT81/HRE-luciferase containsthree tandem copies of the erythropoietin hypoxiaresponsive element (HRE) in front of the herpes simplexthymidine kinase promoter. Twenty-four hours after

transfection cells were treated and assayed for luciferaseactivity, which was measured using an LMax II 384ROM v1.04 reader (Molecular Devices, Sunnyvale, CA,USA) according to the manufacturer’s protocol. Thetested treatments had no effect on the readout of theluciferase reporter assay (assessed by transfection ofpCMV luciferase).

Determination of VEGF and Ang 2 levels

The concentrations of diffusible Ang 2 and VEGF121,165

in the cell culture supernatants were measured byQuantikine ELISA kits (R&D Systems), using monoclonalantibodies directed against human Ang 2 or humanVEGF. Absorbance was measured at 450 nm, with wave-length correction at 570 nm, on a Biotek Synergy HTspectrophotometer (Biotek, Winooski, VT, USA).

MTT cell viability assay

After the treatments, ARPE-19 cells seeded onto 24-well plates were washed twice with PBS and incubatedwith 0.5 mg ml−1 MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Invitrogen], solubilized inKrebs buffer (mM: 130 NaCl, 4 KCl, 1.5 MgCl2, 1 CaCl2,6 glucose and 10 Hepes, pH 7.4), for 2 h at 37◦C in acell culture incubator. Subsequently, supernatants wereremoved, and the precipitated dye was dissolved in300 μl 0.04 M HCl (in isopropanol) and colorimetricallyquantified at 570 nm, with wavelength correction at620 nm, using a Biotek Synergy HT spectrophotometer.

5-Bromo-2′-deoxyuridine (BrdU) incorporationcolorimetric assay

TR-iBRB cells seeded onto a 96-well plate were incubatedwith BrdU labelling solution to a final concentration of10 μM for 6 h at 37◦C. Detection of BrdU incorporatedon DNA during proliferation was assessed using aspecific antibody against BrdU conjugated with peroxidase(anti-BrdU-POD), which develops a photometric signalin the presence of the substrate solution tetramethylbenzidine (TMB), as described by the manufacturer’sprotocol (Roche Applied Science, Indianapolis, IN, USA).Absorbance was measured at 370 nm on a Biotek SynergyHT spectrophotometer.

Caspase-3 cleavage activity colorimetric assay

TR-iBRB cells cultured in 100 mm × 20 mm plates weretreated as described in the legend of the figures.Subsequently, cells were lysed, and protein concentrationwas determined using the BCA method (Pierce-ThermoScientific, Waltham, MA, USA). The protein lysatewas incubated with a specific caspase-3 substratethat is conjugated to the colour reporter molecule



p-nitroanaline (DEVD-pNA) for 2 h at 37◦C, according tothe manufacturer’s protocol (R&D Systems). The cleavageof the peptide was quantified spectrophotometricallyat 405 nm using a Biotek Synergy HT spectrophoto-meter.

Fibrin gel in vitro angiogenesis assay

TR-iBRB cells resuspended in positive control media ortest media were embedded within fibrinogen/thrombinpolymerized gels in 96-well plates at a final concentrationof 5 × 104 cells ml−1, as described by the manufacturer’sprotocol (Biotek). Cells were incubated for 24 or 48 h, andphase-contrast images were taken using an inverted lightmicroscope at 100 and ×200 magnification.

Statistical analysis

Data are reported as the means ± S.D. or S.E.M. of at leastthree independent experiments. Comparisons betweenmultiple groups were performed by one-way ANOVA,while comparison between two groups was performedusing an unpaired Student’s t test (GraphPad Prism 5.0software, La Jolla, CA, USA). In all cases, P < 0.05 wasconsidered significant.

Additional details regarding materials and methods aredescribed in the online Supplemental Methods.

Results

Methylglyoxal destabilized HIF-1α and decreased theHIF-1 transcriptional activity in RPE cells

Cell dysfunction in diabetes is related, at least inpart, to lack of cellular ability to adapt to lowoxygen. For example, Catrina et al. (2004) showed thathyperglycaemia downregulates HIF-1α by a p53- andproline hydroxylation-independent mechanism. Based onprevious data, we hypothesize or hypothesized that MGOis likely to be a molecular link between hyperglycaemia anddestabilization of HIF-1 in diabetes. Indeed, data in thisstudy showed that MGO downregulated HIF-1α proteinlevels (Fig. 1A). Consistently, immunofluorescence datashowed that hypoxia led to a remarkable accumulationof HIF-1α in the nucleus, while MGO induced a markeddecrease in the levels of HIF-1α present in the nucleus(Fig. S1). This MGO-induced decrease in protein levels islikely to account for decreased transcriptional activity ofHIF-1. Indeed, through an HRE-driven luciferase reporterassay, we observed that MGO significantly decreasedthe activation of HIF-1 (Fig. 1B). This indicates thatMGO-induced destabilization of HIF-1α correlates withdecreased transcriptional activity of HIF-1. In this study,we used a retinal epithelial cell line (ARPE-19) to

investigate the role of MGO in regulation of HIF-1. Retinalpigment epithelial cells are a good model to evaluatethe cellular response to hypoxia because these cells areextremely sensitive to variations in oxygen concentration(Yu & Cringle, 2005) and are a major source of pro-and anti-angiogenic factors in the eye (Slomiany &Rosenzweig, 2004; Strauss, 2005). Moreover, incubationof ARPE-19 cells with 3 mM MGO for 90 min or 3 hresulted in an intracellular accumulation of MGO of0.77 ± 0.135 and 0.63 ± 0.113 nmol MGO (mg protein)−1

(versus 0.35 ± 0.074 nmol MGO (mg protein)−1; Fig. 1C).The decrease of MGO after 3 h of incubation probablyreflects the scavenging of MGO by glyoxalases (GLOs)and/or irreversible binding of MGO to proteins and othercell components. Moreover, the concentrations and timepoints used to treat ARPE-19 cells with MGO did notresult in significant cell death in comparison with controlcultures (P > 0.05, one-way ANOVA; Fig. 1D). Althoughthe high concentrations used in this work are not found inthe plasma and tissues, we observed that they induced anintracellular accumulation of MGO that is consistent withthe levels found in tissues of diabetic animals (compareFigs 1C and 5A; Ma et al. 2009). Moreover, the obtainedintracellular levels of MGO using these concentrations aresimilar to those observed in other studies using cell culturesystems (Miller et al. 2006). Based on some reports, it isestimated that only a small fraction of exogenous MGO istaken into cells (Che et al. 1997). Presumably, most of theMGO added to cells reacts with components of the culturemedium (e.g. serum), is retained on the cell surface and/ordetoxified by the GLO system.

Methylglyoxal increased the levels of Ang 2 anddecreased the levels of VEGF secreted by RPE cells

It was recently shown that MGO increases the expressionof Ang 2 by a mechanism dependent on the modificationof the corepressor mSin3A (Yao et al. 2007). Moreover,considering prior data, we anticipated that MGO wouldfurther decrease the levels of VEGF secreted by cells inhypoxic conditions, since VEGF is a target gene of HIF-1.Indeed, through Western blotting we observed that MGOincreased the levels of Ang 2 and decreased the levels ofVEGF (Fig. 2A), thus leading to a dramatic change in thebalance between VEGF and Ang 2. Consistently, MGOincreased the levels of Ang 2 and strongly decreased thelevels of VEGF secreted into the extracellular space byRPE cells in response to hypoxia (Fig. 2B).

Data also showed that both GLO I overexpression usingan efficient adenovirus-based delivery system (Fig. 2C)and GLO I silencing using an miRNA system (Fig. 2D)had remarkable effects on the levels of VEGF andAng 2 secreted into the culture medium. Indeed, GLO Ioverexpression increased the levels of VEGF secretedby RPE cells in the presence of MGO. Conversely,



overexpression of GLO I induced a significant decreasein the levels of Ang 2 secreted by these cells, notonly in the presence of MGO, but also in conditionswhere no MGO was added to the medium (Fig. 2E).In contrast, silencing of GLO I had similar effects tothose observed for the addition of MGO concerning thelevels of VEGF and Ang 2 secreted into the medium.Knocking down GLO I decreased the levels of VEGFand increased the levels of Ang 2 secreted by RPEcells (Fig. 2F). Considering that GLO I is the rate-limiting enzyme involved in detoxification of MGO,this observation provides robust evidence that implicatesMGO in changing the VEGF/Ang 2 ratio secreted by thesecells.

A B

C D

0

0.2

0.4

0.6

0.8

1

Control MGO 90' MGO 3h

MG

O C

on

ce

ntr

ati

on

(nm

ol (m

g p

rote

in)–

1)

**

*

0

20

40

60

80

100

120

Control MGO 90' MGO 3h

Ce

ll V

iab

ilit

y (

%)

0

0.5

1

1.5

2

2.5

3

Control MGO Hypoxia Hypoxia + MGO

Lu

cif

era

se

Ac

tiv

ity

(Fo

ld In

du

cti

on

)

*

#

Figure 1. Methylglyoxal decreases the HIF-1α protein levels and decreases HIF-1 transcriptional activityA, ARPE-19 cells were exposed to hypoxia (2% O2) for 6 h in the absence or in the presence of MGO (3 mM

for 10, 60 or 90 min). The cell lysates were analysed by Western blotting (WB) using anti-HIF-1α and anti-actinmonoclonal antibodies. B, ARPE-19 cells were transiently transfected with the pT81 HRE-luciferase vector andthe cells were subjected to hypoxia (2% O2) for 6 h in the absence or presence of MGO (3 mM for 90 min).Subsequently, the luciferase activity was determined and the values were expressed as multiples of the controlvalue. The results represent the means ± S.D. of at least three independent experiments. ∗P < 0.05, significantlydifferent from control; #P < 0.05, significantly different from hypoxic conditions (one-way ANOVA with Tukey’smultiple comparison test). C, ARPE-19 cells were treated with 3 mM of MGO for 90 min or 3 h. Cells were thenlysed, and the intracellular levels of MGO were determined by HPLC after derivatization with 1,2-diamino-4,5-dimethoxybenzene. The results represent the means ± S.D. of at least three independent experiments. ∗P < 0.05and ∗∗P < 0.01, significantly different from control values (one-way ANOVA). D, ARPE-19 cells were treated with3 mM of MGO for 90 min or 3 h and used to assess cell viability through the MTT colorimetric assay. The quantifiedresults represent the means ± S.D. of at least three independent experiments (P > 0.05, one-way ANOVA, Tukey’spost hoc test).

Preconditioned medium from epithelial cells treatedwith MGO activated apoptosis and decreasedproliferation of endothelial cells

Imbalance in the VEGF/Ang 2 ratio is likely to induceendothelial dysfunction. We hypothesized that thisimbalance induced by MGO in RPE cells would leadto activation of apoptosis and decrease proliferation ofendothelial cells. To evaluate this hypothesis, TR-iBRBcells were treated with preconditioned media obtainedfrom ARPE-19 cells subjected to hypoxia in the presenceor absence of MGO. TR-iBRB cells exhibit properties ofretinal capillary endothelial cells and have been describedas an adequate model of inner blood–retinal barrier(Hosoya et al. 2001).



A

B

C D

E

F

0

500

1000

1500

2000

2500


VE

GF

Co

nc

en

tra

tio

n (

pg

ml–

1)

***

***

###

0

200

400

600

800

1000


An

g 2

Co

nc

en

tra

tio

n (

pg

ml–

1)

*#

0

1

2

3

4

5


Rati

o V

EG

F/A

ng

2

***

***

###

0

300

600

900

1200

1500

1800


VE

GF

Co

nc

en

tra

tio

n (

pg

ml–

1)

Mock

GLO I

*

*

0

200

400

600

800

1000

1200


An

g 2

Co

nc

en

tra

tio

n (

pg

ml–

1)

Mock

GLO I

**

****

**

0

200

400

600

800

1000

Mock miRNA GLO I

VE

GF

Co

nc

en

tra

tio

n (

pg

ml–

1)

*

0

200

400

600

800

Mock miRNA GLO I

An

g 2

Co

nc

en

tra

tio

n (

pg

ml–

1) *

Ang 2

Figure 2. Methylglyoxal disrupts the balance between VEGF and Ang 2 produced and secreted by RPEcellsA, ARPE-19 cells were exposed to hypoxia (2% O2) for 6 h and treated with 3 mM MGO for 90 min. The celllysates were analysed by Western blotting (WB) using anti-VEGF, anti-Ang 2 and anti-actin antibodies. B, ARPE-19cells were subjected to hypoxia (2% O2) for 12 h either in the absence or in the presence of 1 mM MGO. The



Considering that ARPE-19 cells are derived fromhumans (Homo sapiens) and TR-iBRB cells are derivedfrom rats (Rattus norvegicus), the homology of VEGF-Aand Ang 2 between both species was investigated beforeproceeding with additional experiments. As observed inFig. 3A, the sequences of both VEGF-A and Ang 2 arehighly conserved between both species; human VEGF-A(NP 001020537.2) shares 84% of consensus positionswith rat VEGF-A (NP 114024.2), while human Ang 2(NP 001138.1) shares 90.7% of consensus positions withthe rat Ang 2 protein (NP 604449.1). Moreover, datapublished in NCBI databases mention that the VEGF-A binding interface comprises the 230–233, 267, 269–272 and 287–290 residues, while the F468A/Y474A/Y475Aand K467E/K472E Ang 2 mutants do not bind Tie2(Barton et al. 2005). Vascular endothelial growthfactor and Ang 2 sequence alignments show that thementioned residues (underlined in Fig. 3A) are fullyconserved between both species. Thus, this homologystudy excludes incompatibility between human VEGF-A and Ang 2 with rat VEGF-R and Tie2 receptors,respectively.

After the homology studies and the validation of ourpreconditioned cell culture system, the levels of BAXand Bcl-2 in the endothelial cells were determined byimmunoblotting. It is well established that the BAX/Bcl-2ratio determines the response to a death signal (Oltvai et al.1993), being a marker of apoptosis activation. Moreover,induction of BAX expression is known to initiate death inthe absence of any additional signal (Xiang et al. 1996),resulting in a downstream programme of mitochondrialdysfunction, as well as activation of caspases. In thisstudy, we observed that the preconditioned medium ofARPE-19 cells subjected to hypoxia strongly increased theBcl-2 levels in endothelial cells (Fig. 3B). Conversely, the

concentration of diffusible Ang 2 and VEGF-A121/165 isoforms secreted into the culture medium was determinedby ELISA using specific monoclonal antibodies against human VEGF and Ang 2. The results represent themeans ± S.D. of at least three independent experiments. ∗P < 0.05 and ∗∗∗P < 0.001, significantly differentfrom control values; #P < 0.05 and ###P < 0.001, significantly different from hypoxia (one-way ANOVA withTukey’s multiple comparison test). C, ARPE-19 cells were infected with pAd V5-GLOI for 48 h. The cell lysates wereanalysed by immunoblotting using anti-V5 and anti-actin monoclonal antibodies. D, ARPE-19 cells were transfectedwith pcDNA6.2-GW/EmGFP-miRNA GLO I for 48 h. Cells were subsequently lysed, and protein extracts wereimmunoblotted for GLO I and actin. E, ARPE-19 cells were infected with pAd V5-GLOI for 48 h. After infection,the cell culture media were replaced by fresh media and cells were subjected to hypoxia for 12 h either in theabsence or in the presence of 1 mM MGO. The concentration of diffusible Ang 2 and VEGF-A121/165 isoformssecreted into the culture medium was determined by ELISA using specific monoclonal antibodies against humanVEGF and Ang 2. The results represent the means ± S.D. of at least three independent experiments. ∗P < 0.05and ∗∗P < 0.01 (one-way ANOVA with Tukey’s multiple comparison test). F, ARPE-19 cells were transfected withpcDNA6.2-GW/EmGFP-miRNA GLO I for 48 h. After transfection, the cell culture media were replaced by freshmedia, and cells were subjected to hypoxia for 12 h either in the absence or in the presence of 1 mM MGO. Theconcentration of diffusible Ang 2 and VEGF-A121/165 isoforms secreted into the culture medium was determined byELISA using specific monoclonal antibodies against human VEGF and Ang 2. The results represent the means ± S.D.of at least three independent experiments. ∗P < 0.05 (one-way ANOVA with Tukey’s multiple comparison test).The word ‘Mock’ in the figures refers to a control with a scrambled miRNA sequence or with an empty vector,depending on the purpose of the experiment.

preconditioned medium of ARPE-19 cells simultaneouslysubjected to hypoxia and MGO decreased the Bcl-2/BAXratio (Fig. 3B) and increased activation of caspase-3(Fig. 3C), suggesting activation of apoptosis. We furtherobserved that incubation of endothelial cells with thispreconditioned medium decreased the incorporation ofBrdU, a synthetic thymidine analogue, in DNA (Fig. 3D)and decreased the levels of the proliferation markerKi67 (Fig. 3E and F), indicating decreased proliferationof endothelial cells. The proliferation marker Ki67 isknown to be located in the nucleolus during interphase,while it occurs associated with condensed chromosomesduring mitosis (Scholzen & Gerdes, 2000). Indeed, ‘MGO’and ‘Hypoxia + MGO’ conditions were characterizedby a small number of cells with a Ki67 mitosis pattern,suggesting decreased cell division/proliferation.

Angiopoietin 2, in the presence of low levels of VEGF,decreased proliferation and activated apoptosisof endothelial cells

Hypoxia and MGO are likely to change the levels ofseveral growth factors and cytokines released by RPE cells,possibly leading to changes in the BAX/Bcl-2 ratio andcompromising survival of endothelial cells. To specificallyassess the role of VEGF and Ang 2 in the proliferationand survival of endothelial cells, we treated cells withrecombinant VEGF and Ang 2 within the same ratiosof VEGF/Ang 2 as those that were released by ARPE-19 cells following treatment with MGO and/or hypoxia.We observed that Ang 2, in the presence of high levelsof VEGF (conditions that ‘mimic’ hypoxia), stronglydecreased the levels of BAX (Fig. 4A) in endothelial cells.In contrast, we found that Ang 2, in the presence of



201 250hVEGFA (201) AKWSQAAPMAEGGGQNHHEVVKFMDVYQRSYCHPIETLVDIFQEYPDEIErVEGFA (200) AKWSQAAPTTEGE-QKAHEVVKFMDVYQRSYCRPIETLVDIFQEYPDEIE

251 300hVEGFA (251) YIFKPSCVPLMRCGGCCNDEGLECVPTEESNITMQIMRIKPHQGQHIGEMrVEGFA (249) YIFKPSCVPLMRCAGCCNDEALECVPTSESNVTMQIMRIKPHQSQHIGEM

451 496hAng 2 (451) GPSNLNGMYYPQRQNTNKFNGIKWYYWKGSGYSLKATTMMIRPADFrAng 2 (451) GPSNLNGQYYPQKQNTNKFNGIKWYYWKGSGYSLKATTMMIRPADF

A

B

E F

DC

0

0.5

1

1.5

2

2.5

3

3.5

4

Control Hypoxia MGO Hypoxia + MGO

Ra

tio

Bc

l-2

/BA

X

***

** ###

0

0.5

1

1.5

2

2.5


Ca

sp

as

e-3

Ac

tiv

ity

(Fo

ld I

nd

uc

tio

n)

*

0

0.4

0.8

1.2

1.6

2


Brd

U In

co

rpo

rati

on

***###

Figure 3. Preconditioned media of RPE cells treated with MGO activate apoptosis and decreaseproliferation of retinal endothelial cellsA, partial VEGF and Ang 2 sequence alignments between Homo sapiens and Rattus norvegicus species. F, non-similar; M, identical; A, block of similar; and KF, receptor-binding interface. B, ARPE-19 cells were exposed tohypoxia (2% O2) for 12 h either in the absence or in the presence of 1 mM MGO. After the elimination of MGOthat might remain in the media, preconditioned media were used to incubate TR-iBRB cells for 24 h. The celllysates were analysed by Western blotting (WB) using specific antibodies against BAX, Bcl-2 and actin. The resultsrepresent the means ± S.D. of at least three independent experiments. ∗∗P < 0.01 and ∗∗∗P < 0.001, significantly



low levels of VEGF (conditions that ‘mimic’ hypoxia andMGO treatment), induced a strong decrease in the Bcl-2levels (Fig. 4A), dramatically altering the BAX/Bcl-2 ratioand presumably leading to apoptosis. To further confirmthis hypothesis, we determined the proteolytic activityof caspase-3, a key executioner of apoptosis (Cohen,1997; Smulson et al. 1998; Porter & Janicke, 1999), byusing a specific caspase-3 substrate. Data showed that inthe presence of low levels of VEGF, Ang 2 significantlyincreased the caspase-3 activity (Fig. 4B). Consistently,we also observed that these conditions decreased theincorporation of BrdU in DNA (Fig. 4C), indicatingdecreased cell proliferation. In order to further assessthe role of the VEGF/Ang 2 ratio in proliferation ofretinal endothelial cells, we performed a fibrin gel in vitroangiogenic assay to assess the ability of cultured cells toform tube-like structures. When cultured within a fibringel and subjected to angiogenic signals, endothelial cellsrapidly align and form interconnecting networks. By usingseveral ratios of VEGF/Ang 2, we observed that Ang 2, inthe presence of high levels of VEGF, induced accentuatedcell proliferation (Fig. 4D). Moreover, endothelial cellsaligned, forming closed polygons and a network of proto-tubes. However, Ang 2 in the presence of low levels ofVEGF did not favour the formation of proto-tubes orclosed polygons, but rather induced isolated growing cellsand a few emergent tubes of low complexity (Fig. 4D).

The retinas of diabetic rats presented increased levelsof MGO and changes in the expression anddistribution of HIF-1α, VEGF and Ang 2

In this study, we found that MGO destabilized HIF-1α

in hypoxic conditions and imbalanced the VEGF/Ang 2ratio secreted by RPE cells, activating apoptosis anddecreasing proliferation of endothelial cells in culture.In order to assess the relevance of these data in a morephysiological context, we investigated whether some ofthe cardinal MGO-induced modifications in cell culturesystems were also present in retinas of an animal modelof diabetes. For this purpose, we used the GK animalmodel of diabetes (Goto et al. 1976), characterized by

different from control value; ###P < 0.001, significantly different from hypoxia (one-way ANOVA with Tukey’smultiple comparison test). C, TR-iBRB cells, treated with preconditioned media from ARPE-19 cells, were usedto determine the caspase-3 cleavage activity by spectrophotometry. The results represent the means ± S.D. of atleast three independent experiments. ∗P < 0.05, significantly different from control values (one-way ANOVA withTukey’s multiple comparison test). D, TR-iBRB cells, treated with preconditioned media from ARPE-19 cells, wereincubated with a BrdU labelling solution, and detection of BrdU incorporated into DNA was assessed according tothe manufacturers’ instructions. The results represent the means ± S.D. of at least three independent experiments.∗∗∗P < 0.001, significantly different from control values; ###P < 0.001, significantly different from hypoxia (one-way ANOVA with Tukey’s multiple comparison test). E and F, TR-iBRB cells, incubated as previously described,were used to assess the protein levels (E) and the cellular distribution (F) of the proliferation marker Ki67, byimmunoblotting and immunofluorescence, respectively.

increased fasting glycaemia, hyperglycaemia after 2 h ofglucose administration and increased levels of glycatedhaemoglobin (Table S1). The GK rat appears to be anappropriate animal model for the study of retinal changesassociated with diabetes. Indeed, and as observed fordiabetic patients, the GK rats present reduced retinalβ-wave amplitudes and reduced oscillatory potentials.These functional abnormalities of the photoreceptorsmight be associated with inheritable retinal degenerationat an early age (Matsubara et al. 2006). Abnormalitiesin microcirculation have also been observed in GK ratsat 5 months old, before the appearance of ‘clinically’detectable retinal changes (Miyamoto et al. 1996).Additional features characteristic of diabetic retinal diseaseinclude increased apoptosis of retinal microvascular cellsand pericytes after 8 months old (Agardh et al. 1997).

Data showed that the retinas of GK rats had increasedlevels of MGO compared with age-matched controlWistar retinas (Fig. 5A). We further observed importantchanges in the expression and distribution of some ofthe key proteins studied in the cell culture systems.For example, HIF-1α protein was less abundant in theretina of diabetic rats and was mostly expressed in theretinal pigment epithelium and outer plexiform layers(Fig. 5B) of non-diabetic animals. Consistent with ourmodel, data further showed that VEGF expression waslower in retinas of diabetic compared with non-diabeticrats. Indeed, VEGF was highly expressed in the retinalpigment epithelium and ganglion cell layers (Fig. 5C)of Wistar animals. Moreover, data also showed thatAng 2 was more abundant in the retina of GK rats(Fig. 5D). Consistently, we observed by immunoblottingthat retinas of diabetic rats presented decreased levelsof HIF-1α and VEGF and increased levels of Ang 2compared with retinas of non-diabetic animals (Fig. 5E).Taken together, these changes are likely to contribute toa general imbalance between VEGF and Ang 2 acrossthe retina of diabetic animals. Immunohistochemistrydata further showed changes in the distribution andexpression of BAX and Bcl-2 between retinas of diabeticand non-diabetic animals (Fig. 5F and G). For example,GK rats appeared to have low levels of Bcl-2 and highlevels of BAX, which may result in an increased amount



A

B C

D

0

2

4

6

8

10

I II IIIR

ati

o B

cl-

2/B

AX

***

**

0

0.3

0.6

0.9

1.2

1.5

I II III

Ca

sp

as

e-3

Ac

tiv

ity

(F

old

In

du

cti

on

)

*

*

0

5

10

15

20

25

30

I II IIINu

mb

er

of

clo

se

d s

tru

ctu

res

pe

r w

ell

*

0

0.2

0.4

0.6

0.8

1

1.2

1.4

I II III

Brd

U In

co

rpo

rati

on

*

*

Figure 4. In the presence of low levels of VEGF, Ang 2 decreases proliferation and activates the apoptoticcascade in retinal endothelial cellsA, TR-iBRB cells, cultured in DMEM without serum, were incubated with the following concentrations ofrecombinant VEGF165 and Ang 2 for 24 h: I, 6 ng ml−1 VEGF + 4 ng ml−1 Ang 2; II, 20 ng ml−1 VEGF + 4 ng ml−1

Ang 2; and III, 0.5 ng ml−1 VEGF + 7 ng ml−1 Ang 2. Subsequently, the cell lysates were analysed by Westernblotting (WB) against BAX, Bcl-2 and actin. The results represent the means ± S.D. of at least three independentexperiments. ∗∗P < 0.01 and ∗∗∗P < 0.001, significantly different from condition I (one-way ANOVA with Dunnett’smultiple comparison test). B and C, TR-iBRB cells, incubated as previously described, were used to determine thecaspase-3 cleavage activity by spectrophotometry (B) and the proliferation activity through the BrdU incorporation(C). The results represent the means ± S.D. of at least three independent experiments. ∗P < 0.05, significantlydifferent from condition I (one-way ANOVA with Tukey’s multiple comparison test). D, TR-iBRB cells were embeddedwithin fibrinogen/ thrombin polymerized gels and incubated with recombinant VEGF and Ang 2 proteins. Phase-contrast images were taken using an inverted light microscope 48 h following embedding. The results representthe means ± S.D. of at least three independent experiments, where the number of closed structures were countedper well. ∗P < 0.05, significantly different from condition I (one-way ANOVA with Tukey’s multiple comparisontest).



of apoptotic endothelial cells and, consequently, vesselregression, which would translate to less staining for vWF,a specific marker of endothelial cells (Fig. 5H).

In vivo staining of retinal microvessels with Evans Bluefurther showed increased vessel permeability in the retinasof GK rats compared with Wistar rats (Fig. 5I). Evans Bluedye is known to combine with plasma albumin and is awell-accepted tool to monitor vessel leakage. In the earlystages of DR, increased capillary leakage is a main featureof blood–retinal barrier breakdown and is a consequenceof increased capillary fenestration by disruption of tightjunctions between endothelial cells (Harhaj & Antonetti,2004).

Altogether, data obtained in animals are consistent within vitro cell culture data and support a model in whichincreased levels of MGO decrease the ratio between VEGFand Ang 2 released by active secretory cells, such as RPEcells, leading to apoptosis and decreased proliferation ofneighbouring endothelial cells.

Discussion

Hyperglycaemia appears to be a critical factor in theaetiology of diabetes and initiates a number of downstreamevents, such as pericyte drop-out, vascular obstructionand capillary non-perfusion (Waltenberger, 2001), whichculminate in tissue ischaemia. The cellular response tohypoxia is a fairly complex process that involves activationof HIF-1, a critical step to help cells and tissues to surviveand adapt to low oxygen. However, hyperglycaemia wasshown to disrupt, at least in part, the ability of tissues torespond and adapt to hypoxia (Abaci et al. 1999; Malmberget al. 1999; Catrina et al. 2004; Kido et al. 2005). Althoughthe underlying molecular mechanisms are still unclear,it is conceivable that destabilization of HIF-1α by highglucose concentrations is at the origin of the loss of thecellular response to hypoxia in diabetes (Catrina et al.2004). Consistently, downregulation of HIF-1 in responseto hyperglycaemia is likely to account for the decreasedarteriogenic response triggered by myocardial ischaemiain diabetic patients (Abaci et al. 1999; Larger et al. 2004).In addition, levels of HIF-1 are downregulated in biopsiesfrom ulcers of diabetic patients compared with venousulcers that share the same hypoxic environment but arenot exposed to hyperglycaemia (Catrina et al. 2004).These and other lines of evidence strongly suggest thatcellular and tissue dysfunction associated with diabetesis related, at least in part, to the lack of cellular abilityto adapt and survive in hypoxic conditions. Recent datafrom our group suggest that increased production ofMGO is likely to be a link between hyperglycaemia anddestabilization of HIF-1α. Indeed, in the present studywe showed that MGO decreased the levels of HIF-1α

protein during hypoxia, particularly in the nucleus, and

decreased its transcriptional activity. This decrease is likelyto account for the downregulation of VEGF, a well-established prosurvival and pro-angiogenic factor.

Several lines of evidence suggest that an imbalancein the levels of cytokines might be associated withthe vascular complications of diabetes and impairedangiogenic response. Vascularization is regulated bycomplex context-dependent interactions between, amongother factors, Ang 2 and VEGF. Changes in the regulationof either factor are likely to be involved in alterationsof the angiogenic remodelling programme. Diabetesinduces a significant increase in retinal expression ofAng 2 in rats (Hammes et al. 2004), and diabeticAng 2+/− mice have both decreased pericyte loss andreduced acellular capillary formation compared withdiabetic wild-type strain Ang 2+/+ (Hammes et al. 2002).Moreover, experimental evidence suggests that decreasedlevels of VEGF might be associated with wound-healingdisorders in diabetic patients (Altavilla et al. 2001),and simvastatin-induced VEGF expression or topicalVEGF application ameliorates impaired wound healing(Galiano et al. 2004; Singh et al. 2007; Bitto et al. 2008).Furthermore, following an ischaemic insult, VEGF levelsin diabetic hearts remain significantly lower than thosein non-diabetic hearts (Marfella et al. 2002, 2004). Inaddition, myocardial ischaemia-induced VEGF expressionis significantly impaired in diabetic patients (Tuo et al.2008), and some of these changes can be reversedby decreasing the levels of Ang 2. These and otherobservations highlight the critical role for the VEGF–Ang 2balance in the regulation of angiogenesis, in myocardialinjury (Tuo et al. 2008) and in other pathophysiologicalcomplications associated with diabetes.

Accumulation of MGO has been implicated in thedevelopment of many vascular complications of diabetes(Kalapos, 1999). Significantly, increased retinal levels ofMGO in diabetic rats are associated with the formation ofacellular capillaries, drop-out of pericytes (Hammes, 2003;Hammes et al. 2003) and deregulation of angiogenesis(Berlanga et al. 2005), which are major structural lesionsof DR. For example, MGO was shown to increasethe expression of Ang 2, a key destabilizing factorinvolved in the initiation of angiogenic remodelling (Yaoet al. 2007). This mechanism was shown to involveMGO-induced modification of mSin3A that results inincreased recruitment of O-GlcNAc-transferase, with aconsequent increase in modification of Sp3 by O-linkedN-acetylglucosamine. This modification of Sp3 causesdecreased binding to a glucose-responsive GC box in theAng 2 promoter, resulting in increased Ang 2 expression(Yao et al. 2007). Consistently, in this work we showed thatMGO increased the levels of Ang 2 produced and secretedby retinal epithelial cells and decreased the levels of VEGFsecreted by these cells. The resulting imbalance in theVEGF/Ang 2 ratio led to increased endothelial cell death



Figure 5. Goto–Kakizaki retinas have increased levels of MGO and changes in the expression anddistribution of VEGF and Ang 2 compared with Wistar retinasA, Goto–Kakizaki (GK) and Wistar (WS) retinas were lysed and used to determine the levels of MGO by HPLCafter derivatization with 1,2-diamino-4,5-dimethoxybenzene. The results represent the means ± S.D. and are



and decreased proliferation, which are likely to accountfor vascular regression and deregulation of the angiogenicresponse in diabetes. However, one cannot exclude therole of MGO in the deregulation of other angiogenicfactors [such as transforming growth factor-β, tumournecrosis factor-α, interleukins and PEDF (pigmentepithelium-derived factor)] that might exert criticaleffects on the proliferation and apoptosis of endothelialcells.

Although the major purpose of this work was notto study the downstream pathways activated by VEGFand Ang 2 leading to survival, proliferation or apoptosis,it is certainly an interesting issue to pursue in thefuture. Several reports have shown that activation ofVEGF receptors by VEGF-A leads to phosphoinositide 3-kinase, protein kinase B and protein kinase C activation,culminating in increased expression of several survivaland proliferation factors [e.g. survivin, IAP (inhibitorof apoptosis) and Bcl-2] and decreased expressionand/or activity of various apoptotic factors [e.g. BAD(BCL2-associated agonist of cell death), caspases], whileinactivation of the Tie2 receptor by Ang 2 leads toactivation of nuclear factor-κB and p53 and inactivationof PI3K/AKT, leading to increased apoptotic cell death(Petrova et al. 1999; Eklund & Olsen, 2006). Thus, anincrease in the levels of Ang 2 and a decrease in thelevels of VEGF (as induced by MGO) will most probablyinduce activation of apoptotic pathways, in detriment toinactivation of survival and proliferative pathways, leadingto the overall effects reported in this study.

Diabetes has been characterized by an ‘angiogenicparadox’, since both pro- and anti-angiogenic conditionscoexist in chronic diabetes. In some microvasculartissues, increased expression of growth factors results inneovascularization. This response has a significant role inadvanced stages of DR, which seems to be amelioratedby inhibition of angiogenesis. Conversely, decreasedexpression of pro-angiogenic factors is associated withimpairment of new blood vessel formation, leading topoor wound healing, coronary artery and peripheral limbdisease, which is ameliorated by induction of release of

representative of each group, composed of at least three animals. ∗P < 0.05, significantly different from controlvalues (Student’s unpaired t test). In B, C and D, Goto–Kakizaki and Wistar rat retinas were frozen in OCT tissueembedding matrix and subsequently cryosectioned. Retinal slices 5–7 μm thick were fixed with cold acetoneand used to perform immunohistochemistry using specific antibodies against HIF-1α (B), VEGF (C) and Ang 2(D). Retinas were imaged by fluorescence microscopy, and images are representative of each group of animals.E, Wistar and GK retinal extracts were immunoblotted against HIF-1α, VEGF, Ang 2 and actin. Western blots (WB)are representative of each group, composed of at least three animals. The results represent the means ± S.D.;∗P < 0.05, significantly different from Wistar (Student’s unpaired t test). F, G and H, Goto–Kakizaki and Wistarretinas were frozen in OCT tissue embedding matrix and subsequently cryosectioned. Retinal slices 5–7 μm thickwere fixed with cold acetone and used to perform immunohistochemistry using specific antibodies against BAX(F), Bcl-2 (G) and vWF (H). Retinas were imaged by fluorescence microscopy, and images are representative of eachgroup of animals. I, Goto–Kakizaki and Wistar rats were injected with an Evans Blue (red) solution through thejugular vein 30 min before the animals were killed. Flat-mounted retinas were imaged by fluorescence microscopy,and images are representative of each group of animals. The arrows indicate leakage sites.

growth factors. Thus, two apparently opposite angiogenicresponses occur in diabetes and generate an intriguingparadox (Duh & Aiello, 1999).

Several reports suggest that VEGF plays a central rolein mediating diabetic vascular complications and is likelyto be at the basis of the angiogenic paradox of diabetes.In the case of DR, it is clearly established that VEGF isa major mediator of retinal neovascularization, servinga pivotal role in the aetiology of advanced stages of DR.In fact, VEGF concentrations are markedly elevated inboth vitreous and aqueous fluids of patients with activeproliferative DR compared with samples from patientswithout diabetes, with non-proliferative DR or withquiescent proliferative DR (Aiello et al. 1994; Duh & Aiello,1999). However, early stages of DR are characterized byopposite phenotypes, mainly associated with formationof acellular capillaries, pericyte drop-out, impairment ofthe angiogenic response and increased vessel permeability,accounting for blood–retinal barrier breakdown.

Despite all the advances in the understanding of DR,the nature and time course of molecular changes in thediabetic retina are poorly characterized, and studies onexpression of VEGF have produced conflicting results. Forexample, retinal VEGF mRNA levels have been reportedby independent studies to be unchanged at 3 months,increased at 6 months and decreased at 6 months followingstreptozotocin-induced diabetes in rats (Brucklacher et al.2008). Moreover, a recent study, based on a geneexpression profile, shows a consistent decrease in VEGFmRNA expression in the retina of 1- and 3-month-oldstreptozotocin-induced diabetic rats (Brucklacher et al.2008). This work also shows that the decrease in theVEGF mRNA levels is accompanied by increased vascularpermeability and caspase-3 activity, two main features ofthe early stages of DR, which are typically associated withincreased expression of VEGF. This observation was alsocorroborated by the present study. Indeed, in this work weshowed that increased vessel permeability and apoptosiswere accompanied by decreased levels of VEGF in6-month-old GK diabetic rats. Thus, it seems that there aremultiple phases in the regulation of vegf gene expression



at different stages of DR, and it seems imprudent togeneralize the upregulation of VEGF for all the stages ofthe disease.

In conclusion, the results reported in this paper areconsistent with a number of diverse observations thatimplicate the VEGF–Ang 2 interplay in the vascularcomplications observed in diabetes and, in particular, onthe propensity to develop DR. In this study, we suggestthat high glucose, by increasing the intracellular levelsof MGO, is likely to disrupt the VEGF–Ang 2 balance,resulting, at least in part, in characteristic features of theearly stages of diabetic vascular complications, such asendothelial cell death, vascular regression and increasedmicrovascular leakage.

References

Abaci A, Oguzhan A, Kahraman S, Eryol NK, Unal S, Arinc H& Ergin A (1999). Effect of diabetes mellitus on formation ofcoronary collateral vessels. Circulation 99, 2239–2242.

Agardh CD, Agardh E, Zhang H & Ostenson CG (1997).Altered endothelial/pericyte ratio in Goto-Kakizaki ratretina. J Diabetes Complications 11, 158–162.

Aiello LP, Avery RL, Arrigg PG, Keyt BA, Jampel HD, Shah ST,Pasquale LR, Thieme H, Iwamoto MA, Park JE, Nguyen HV,Aiello LM, Ferrara N & King GL (1994). Vascular endothelialgrowth factor in ocular fluid of patients with diabeticretinopathy and other retinal disorders. N Engl J Med 331,1480–1487.

Altavilla D, Saitta A, Cucinotta D, Galeano M, Deodato B,Colonna M, Torre V, Russo G, Sardella A, Urna G, CampoGM, Cavallari V, Squadrito G & Squadrito F (2001).Inhibition of lipid peroxidation restores impaired vascularendothelial growth factor expression and stimulates woundhealing and angiogenesis in the genetically diabetic mouse.Diabetes 50, 667–674.

Barton WA, Tzvetkova D & Nikolov DB (2005). Structure ofthe angiopoietin-2 receptor binding domain andidentification of surfaces involved in Tie2 recognition.Structure 13, 825–832.

Berlanga J, Cibrian D, Guillen I, Freyre F, Alba JS, Lopez-SauraP, Merino N, Aldama A, Quintela AM, Triana ME,Montequin JF, Ajamieh H, Urquiza D, Ahmed N &Thornalley PJ (2005). Methylglyoxal administration inducesdiabetes-like microvascular changes and perturbs thehealing process of cutaneous wounds. Clin Sci (Lond) 109,83–95.

Bitto A, Minutoli L, Altavilla D, Polito F, Fiumara T, Marini H,Galeano M, Calo M, Lo Cascio P, Bonaiuto M, Migliorato A,Caputi AP & Squadrito F (2008). Simvastatin enhancesVEGF production and ameliorates impaired wound healingin experimental diabetes. Pharmacol Res 57, 159–169.

Brucklacher RM, Patel KM, VanGuilder HD, Bixler GV, BarberAJ, Antonetti DA, Lin CM, LaNoue KF, Gardner TW,Bronson SK & Freeman WM (2008). Whole genomeassessment of the retinal response to diabetes reveals aprogressive neurovascular inflammatory response. BMC MedGenomics 1, 26.

Cai J & Boulton M (2002). The pathogenesis of diabeticretinopathy: old concepts and new questions. Eye 16,242–260.

Catrina SB, Okamoto K, Pereira T, Brismar K & Poellinger L(2004). Hyperglycemia regulates hypoxia-inducible factor-1α

protein stability and function. Diabetes 53, 3226–3232.Che W, Asahi M, Takahashi M, Kaneto H, Okado A,

Higashiyama S & Taniguchi N (1997). Selective induction ofheparin-binding epidermal growth factor-like growth factorby methylglyoxal and 3-deoxyglucosone in rat aortic smoothmuscle cells. The involvement of reactive oxygen speciesformation and a possible implication for atherogenesis indiabetes. J Biol Chem 272, 18453–18459.

Cohen GM (1997). Caspases: the executioners of apoptosis.Biochem J 326, 1–16.

Cooper ME, Gilbert RE & Jerums G (1997). Diabeticvascular complications. Clin Exp Pharmacol Physiol 24,770–775.

Duh E & Aiello LP (1999). Vascular endothelial growth factorand diabetes: the agonist versus antagonist paradox. Diabetes48, 1899–1906.

Eklund L & Olsen BR (2006). Tie receptors and theirangiopoietin ligands are context-dependent regulators ofvascular remodeling. Exp Cell Res 312, 630–641.

Engerman RL (1989). Pathogenesis of diabetic retinopathy.Diabetes 38, 1203–1206.

Feng Y, Pfister F, Schreiter K, Wang Y, Stock O, Vom Hagen F,Wolburg H, Hoffmann S, Deutsch U & Hammes HP (2008).Angiopoietin-2 deficiency decelerates age-dependentvascular changes in the mouse retina. Cell Physiol Biochem21, 129–136.

Galiano RD, Tepper OM, Pelo CR, Bhatt KA, Callaghan M,Bastidas N, Bunting S, Steinmetz HG & Gurtner GC (2004).Topical vascular endothelial growth factor acceleratesdiabetic wound healing through increased angiogenesis andby mobilizing and recruiting bone marrow-derived cells. AmJ Pathol 164, 1935–1947.

Goto Y, Kakizaki M & Masaki N (1976). Production ofspontaneous diabetic rats by repetition of selective breeding.Tohoku J Exp Med 119, 85–90.

Hammes HP (2003). Pathophysiological mechanisms ofdiabetic angiopathy. J Diabetes Complications 17, 16–19.

Hammes HP, Du X, Edelstein D, Taguchi T, Matsumura T, JuQ, Lin J, Bierhaus A, Nawroth P, Hannak D, Neumaier M,Bergfeld R, Giardino I & Brownlee M (2003). Benfotiamineblocks three major pathways of hyperglycemic damage andprevents experimental diabetic retinopathy. Nat Med 9,294–299.

Hammes HP, Lin J, Renner O, Shani M, Lundqvist A, BetsholtzC, Brownlee M & Deutsch U (2002). Pericytes and thepathogenesis of diabetic retinopathy. Diabetes 51,3107–3112.

Hammes HP, Lin J, Wagner P, Feng Y, Vom Hagen F, KrzizokT, Renner O, Breier G, Brownlee M & Deutsch U (2004).Angiopoietin-2 causes pericyte dropout in the normal retina:evidence for involvement in diabetic retinopathy. Diabetes53, 1104–1110.

Harhaj NS & Antonetti DA (2004). Regulation of tightjunctions and loss of barrier function in pathophysiology. IntJ Biochem Cell Biol 36, 1206–1237.



Holash J, Maisonpierre PC, Compton D, Boland P, AlexanderCR, Zagzag D, Yancopoulos GD & Wiegand SJ (1999).Vessel cooption, regression, and growth in tumorsmediated by angiopoietins and VEGF. Science 284,1994–1998.

Hosoya K, Tomi M, Ohtsuki S, Takanaga H, Ueda M, Yanai N,Obinata M & Terasaki T (2001). Conditionally immortalizedretinal capillary endothelial cell lines (TR-iBRB) expressingdifferentiated endothelial cell functions derived from atransgenic rat. Exp Eye Res 72, 163–172.

Kalapos MP (1999). Methylglyoxal in living organisms:chemistry, biochemistry, toxicology and biologicalimplications. Toxicol Lett 110, 145–175.

Kido M, Du L, Sullivan CC, Li X, Deutsch R, Jamieson SW &Thistlethwaite PA (2005). Hypoxia-inducible factor 1-alphareduces infarction and attenuates progression of cardiacdysfunction after myocardial infarction in the mouse. J AmColl Cardiol 46, 2116–2124.

Larger E, Marre M, Corvol P & Gasc JM (2004).Hyperglycemia-induced defects in angiogenesis in thechicken chorioallantoic membrane model. Diabetes 53,752–761.

Ma H, Li SY, Xu P, Babcock SA, Dolence EK, Brownlee M, Li J& Ren J (2009). Advanced glycation endproduct (AGE)accumulation and AGE receptor (RAGE) up-regulationcontribute to the onset of diabetic cardiomyopathy. J CellMol Med 13, 1751–1764.

Maisonpierre PC, Suri C, Jones PF, Bartunkova S, Wiegand SJ,Radziejewski C, Compton D, McClain J, Aldrich TH,Papadopoulos N, Daly TJ, Davis S, Sato TN & YancopoulosGD (1997). Angiopoietin-2, a natural antagonist for Tie2that disrupts in vivo angiogenesis. Science 277,55–60.

Malmberg K, Norhammar A, Wedel H & Ryden L (1999).Glycometabolic state at admission: important risk marker ofmortality in conventionally treated patients with diabetesmellitus and acute myocardial infarction: long-term resultsfrom the Diabetes and Insulin-Glucose Infusion in AcuteMyocardial Infarction (DIGAMI) study. Circulation 99,2626–2632.

Marfella R, D’Amico M, Di Filippo C, Piegari E, Nappo F,Esposito K, Berrino L, Rossi F & Giugliano D (2002).Myocardial infarction in diabetic rats: role of hyperglycaemiaon infarct size and early expression of hypoxia-induciblefactor 1. Diabetologia 45, 1172–1181.

Marfella R, Esposito K, Nappo F, Siniscalchi M, Sasso FC,Portoghese M, Di Marino MP, Baldi A, Cuzzocrea S, DiFilippo C, Barboso G, Baldi F, Rossi F, D’Amico M &Giugliano D (2004). Expression of angiogenic factors duringacute coronary syndromes in human type 2 diabetes.Diabetes 53, 2383–2391.

Matsubara H, Kuze M, Sasoh M, Ma N, Furuta M & Uji Y(2006). Time-dependent course of electroretinograms in thespontaneous diabetic Goto-Kakizaki rat. Jpn J Ophthalmol50, 211–216.

Miller AG, Smith DG, Bhat M & Nagaraj RH (2006).Glyoxalase I is critical for human retinal capillary pericytesurvival under hyperglycemic conditions. J Biol Chem 281,11864–11871.

Miyamoto K, Ogura Y, Nishiwaki H, Matsuda N, Honda Y,Kato S, Ishida H & Seino Y (1996). Evaluation of retinalmicrocirculatory alterations in the Goto-Kakizaki rat. Aspontaneous model of non-insulin-dependent diabetes.Invest Ophthalmol Vis Sci 37, 898–905.

Oltvai ZN, Milliman CL & Korsmeyer SJ (1993). Bcl-2heterodimerizes in vivo with a conserved homolog, Bax,that accelerates programmed cell death. Cell 74,609–619.

Oshima Y, Deering T, Oshima S, Nambu H, Reddy PS, KalekoM, Connelly S, Hackett SF & Campochiaro PA (2004).Angiopoietin-2 enhances retinal vessel sensitivity tovascular endothelial growth factor. J Cell Physiol 199,412–417.

Petrova TV, Makinen T & Alitalo K (1999). Signaling viavascular endothelial growth factor receptors. Exp Cell Res253, 117–130.

Porter AG & Janicke RU (1999). Emerging roles of caspase-3 inapoptosis. Cell Death Differ 6, 99–104.

Ramasamy R, Yan SF & Schmidt AM (2006). Methylglyoxalcomes of AGE. Cell 124, 258–260.

Scholzen T & Gerdes J (2000). The Ki-67 protein: from theknown and the unknown. J Cell Physiol 182, 311–322.

Shoji T, Koyama H, Morioka T, Tanaka S, Kizu A, MotoyamaK, Mori K, Fukumoto S, Shioi A, Shimogaito N, Takeuchi M,Yamamoto Y, Yonekura H, Yamamoto H & Nishizawa Y(2006). Receptor for advanced glycation end products isinvolved in impaired angiogenic response in diabetes.Diabetes 55, 2245–2255.

Singh AK, Gudehithlu KP, Patri S, Litbarg NO, Sethupathi P,Arruda JA & Dunea G (2007). Impaired integration ofendothelial progenitor cells in capillaries of diabetic woundsis reversible with vascular endothelial growth factor infusion.Transl Res 149, 282–291.

Slomiany MG & Rosenzweig SA (2004). IGF-1-induced VEGFand IGFBP-3 secretion correlates with increased HIF-1α

expression and activity in retinal pigment epithelial cell lineD407. Invest Ophthalmol Vis Sci 45, 2838–2847.

Smulson ME, Pang D, Jung M, Dimtchev A, Chasovskikh S,Spoonde A, Simbulan-Rosenthal C, Rosenthal D, Yakovlev A& Dritschilo A (1998). Irreversible binding ofpoly(ADP)ribose polymerase cleavage product to DNA endsrevealed by atomic force microscopy: possible role inapoptosis. Cancer Res 58, 3495–3498.

Stitt AW, McGoldrick C, Rice-McCaldin A, McCance DR,Glenn JV, Hsu DK, Liu FT, Thorpe SR & Gardiner TA(2005). Impaired retinal angiogenesis in diabetes: role ofadvanced glycation end products and galectin-3. Diabetes 54,785–794.

Strauss O (2005). The retinal pigment epithelium in visualfunction. Physiol Rev 85, 845–881.

Tamarat R, Silvestre JS, Huijberts M, Benessiano J, EbrahimianTG, Duriez M, Wautier MP, Wautier JL & Levy BI (2003).Blockade of advanced glycation end-product formationrestores ischemia-induced angiogenesis in diabetic mice.Proc Natl Acad Sci U S A 100, 8555–8560.

Teixeira AS & Andrade SP (1999). Glucose-induced inhibitionof angiogenesis in the rat sponge granuloma is prevented byaminoguanidine. Life Sci 64, 655–662.



Tuo QH, Zeng H, Stinnett A, Yu H, Aschner JL, Liao DF &Chen JX (2008). Critical role of angiopoietins/Tie-2 inhyperglycemic exacerbation of myocardial infarction andimpaired angiogenesis. Am J Physiol Heart Circ Physiol 294,H2547–2557.

Waltenberger J (2001). Impaired collateral vessel developmentin diabetes: potential cellular mechanisms and therapeuticimplications. Cardiovasc Res 49, 554–560.

Xiang J, Chao DT & Korsmeyer SJ (1996). BAX-induced celldeath may not require interleukin 1β-convertingenzyme-like proteases. Proc Natl Acad Sci U S A 93,14559–14563.

Yancopoulos GD, Davis S, Gale NW, Rudge JS, Wiegand SJ &Holash J (2000). Vascular-specific growth factors and bloodvessel formation. Nature 407, 242–248.

Yao D, Taguchi T, Matsumura T, Pestell R, Edelstein D,Giardino I, Suske G, Rabbani N, Thornalley PJ, Sarthy VP,Hammes HP & Brownlee M (2007). High glucose increasesangiopoietin-2 transcription in microvascular endothelialcells through methylglyoxal modification of mSin3A. J BiolChem 282, 31038–31045.

Yu DY & Cringle SJ (2005). Retinal degeneration and localoxygen metabolism. Exp Eye Res 80, 745–751.

Acknowledgements

This work was supported by the PhD fellowshipSFRH/BD/15229/2004 (C. F. Bento) and the grants POCTI/SAU-OBS/57772/2004 and PDTC/SAU-OSM/67498/2006 (FCT –Fundacao para a Ciencia e a Tecnologia, Portugal).

Supporting Information

Online supplemental material for this paper can be accessed at:

http://jp.physoc.org/cgi/content/full/jphysiol.2008.165084/DC1

As a service to our authors and readers, this journal providessupporting information supplied by the authors. Such materialsare peer-reviewed and may be re-organized for online delivery,but are not copy-edited or typeset. Technical support issuesarising from supporting information (other than missing files)should be addressed to the authors.


methylglyoxalinduced imbalance in the ratio of vascular … · exp physiol 00.00 pp 1–16 1...

Documents