Biological assays and genomic analysis reveal lipoic acid ...

Carcinogenesis vol.28 no.5 pp.1008–1020, 2007doi:10.1093/carcin/bgl233Advance Access publication November 24, 2006

Biological assays and genomic analysis reveal lipoic acid modulation of endothelialcell behavior and gene expression

Patrizia Larghero, Roberta Vene1, Simona Minghelli,Giorgia Travaini, Monica Morini1, Nicoletta Ferrari1,Ulrich Pfeffer1, Douglas M.Noonan2, Adriana Albini3 andRoberto Benelli1

Centro di Biotecnologie Avanzate, Genova, Italy, 1Servizio di OncologiaSperimentale-A Istituto Nazionale per la Ricerca sul Cancro, Genova, Italy,2Dipartimento di Scienze Cliniche e Biologiche, Universita dell’Insubria,Varese, Italy and 3Polo Scientifico e Tecnologico, IRCCS Multimedica,Via Fantoli, 16/15, 20100 Milano, Italy

�To whom correspondence should be addressed. Tel: þ39 02 55406574;Fax: þ39 101 5737231;Email: [email protected]

Lipoic acid (LA) is a sulfated antioxidant produced physiologi-cally as a coenzyme of the pyruvate dehydrogenase complex; it iscurrently used for treatment of non-insulin-dependent diabetes tofavor the cellular uptake of glucose. We have previously describedthe angiopreventive potential of molecules sharing common fea-tures with LA: N-acetyl cysteine, epigallocatechin-3-gallate andxanthohumol. To expand these studies, we have tested the capacityof LA to modulate angiogenesis in tumor growth using a Kaposi’ssarcoma model. Endothelial cells exposed to LA displayed a dose-dependent reduction of cell migration and a time-dependent mod-ulation of the phosphorylation of key signaling molecules. In vivo,LA efficiently repressed angiogenesis in matrigel plugs andKS-Imm tumor growth. We analyzed modulation of gene expres-sion in endothelial cells treated with LA for 5 h (early response),finding a mild anti-apoptotic, antioxidant and anti-inflammatoryresponse. A group of LA-targeted genes was selected to performreal-time polymerase chain reaction time-lapse experiments. Thelong-term gene regulation (48 h and 4 days) shows higher rates ofmodulation as compared with the array data, confirming that LA isable to switch the regulation of several genes linked to cell survival,inflammation and oxidative stress. LA induced the production oftumor necrosis factor-alpha-related apoptosis-inducing ligand(TRAIL) in KS-Imm and activin-A in KS-Imm and endothelialcells; these factors show anti-angiogenic activity in vivo contrib-uting to explain the inhibitory effect of LA on neovascularization.According to our data, LA has promising anti-angiogenic prop-erties, though its influence on central metabolic pathways shouldsuggest more caution about its widespread and not prescribed useat pharmacological doses.

Introduction

Lipoic acid (LA), also known as thioctic acid, 1,2-dithiolane-3-pen-tanoic acid and 1,2-dithiolane-3 valeric acid, is a naturally occurringcompound that is synthesized by plants and animals. LA is not foundin our bodies as a detectable, free circulating form, but bound toa lysine residue in proteins. In its protein bound form, lipoamide,LA is a required cofactor for several multi-enzymatic complexes thatcatalyze critical energy metabolism reactions.

Early studies evaluated the effect of low-dose LA on lipid and car-bohydrate catabolism, observing little or no effect (1); yet, a commonresponse in these trials was an increased glucose uptake from serum

(2). LA increases glucose uptake through recruitment of the glucosetransporter-4 to plasma membranes, a mechanism that is shared withinsulin-stimulated glucose uptake (3). LA improves glucose clearancein patients with type-II diabetes (4) and is particularly suited to the pre-vention of diabetic complications that arise from an overproduction ofreactive oxygen and nitrogen species, due to its antioxidant properties.In experimental and clinical studies, LA markedly reduced the symp-toms of diabetic pathologies, including polyneuropathy and vasculardamage (5,6). In human trials, LA was usually tested for 3–6 weeks atpharmacological oral doses of 600–1200 mg/day (6,7). ExogenousLA is readily and almost completely absorbed, with a limited absolutebioavailability of �30%, caused by high hepatic extraction (8).

Zhang et al. (9) showed that pre-incubation of human aortic endo-thelial cells for 48 h with LA (0.05–1 mM) dose dependently inhibitedtumor necrosis factor (TNF)-alpha-induced expression of E-selectin,vascular cell adhesion molecule 1, intercellular adhesion molecule1 and macrophage chemotactic protein 1, without affecting the ex-pression of TNF-alpha receptor 1. LA dose dependently inhibitedTNF-alpha-induced IkB kinase activation and nuclear translocationof nuclear factor-kappaB (NF-kB). This observation suggested a pos-sible application of LA to block the inflammatory stimuli actingon the endothelial cell in tumor micro-environment. Tumor develop-ment is a multi-step process where angiogenesis is fundamental fromthe beginning, and the early imbalance of the angiogenic switch to-wards inhibition could induce tumor dormancy and significantly pre-vent cancer progression (10). Most chemopreventive molecules exerta powerful anti-inflammatory and antioxidant activity, counteractingpart of the epigenetic signals involved in tumor development (11). Wehave previously described the angiopreventive potential of a moleculesharing common features with LA, N-acetyl cysteine (NAC). NAC, inaddition to its antioxidant and glutathione-restoring activity is able toinhibit tumor and endothelial cell invasion in vitro and to restrain thevascularization of matrigel plugs containing angiogenic growth fac-tors in vivo (12,13). In addition, we have shown that other antioxidantflavonoids inhibit the NF-kB and Akt pathways, including the greentea polyphenol epigallocatechin-3-gallate (EGCG) (14) and the hops-derived chalchone xanthohumol (15). Since previous studies sug-gested that LA could target these same pathways, we tested LA insimilar experimental settings.

Materials and methods

Reagents

All growth factors (acidic fibroblast growth factor (aFGF), basic fibroblastgrowth factor (bFGF), epithelial growth factor (EGF)) were purchased fromPeproTech (London, UK), heparin was purchased from ICN (Irvine, CA),water-soluble hydrocortisone and inhibin-ba/activin-A were purchased fromSigma (St Louis, MO), LA was supplied by Antibioticos (Milan, Italy), tumornecrosis factor-alpha-related apoptosis-inducing ligand (TRAIL) was pur-chased from Alexis Biochemicals (San Diego, CA). Kaposi’s sarcoma cell-conditioned medium (KS-CM) and NIH 3T3-conditioned medium wereobtained as described previously (16,17).

Cell lines

The Kaposi’s sarcoma cell line, KS-Imm, established in our laboratory (18),was cultured in Dulbecco’s modified Eagle’s medium 10% fetal calf serumsplitting the cells 1:3 every other day. Primary human umbilical vein endothe-lial cells (HUVEC) cell cultures were purchased from Cascade Biologics (Port-land, OR) and cultured in gelatin-coated flasks using M199, 10% fetal calfserum supplemented with FGF (1 lg aFGF þ 1 lg bFGF/100 ml medium),EGF (1 lg/100 ml medium), heparin (10 mg/100 ml medium) and hydrocor-tisone (0.1 mg/100 ml medium).

In vivo angiogenesis

The matrigel model of angiogenesis in vivo introduced by Passaniti et al. (19)and modified by Albini et al. (20) was utilized. KS-CM was concentrated

Abbreviations: EGCG, epigallocatechin-3-gallate; GPCR, G protein-coupledreceptor; HO-1, heme oxygenase-1; KS-CM, Kaposi’s sarcoma cell-conditioned medium; LA, lipoic acid; NAC, N-acetyl cysteine; NF-kB, nuclearfactor-kappaB; PBS, phosphate-buffered saline; PCR, polymerase chain reac-tion; TBS, Tris-buffered saline; TNF, tumor necrosis factor; TRAIL, tumornecrosis factor-alpha-related apoptosis-inducing ligand; VEGF, vascular endo-thelial growth factor.

� The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected] 1008

Downloaded from https://academic.oup.com/carcin/article-abstract/28/5/1008/2476405by gueston 13 April 2018

10� in Centricon (cut-off 3 kDa, Amicon-Millipore, Billerica MA) and addedto liquid matrigel containing heparin at 4�C to a final volume of 0.5 ml. C57/bl6 mice were treated everyday with LA (86 lg per mouse per day, correspond-ing to �150 mg/kg dose) in drinking water, starting from 3 days before ma-trigel injection. Ten implants were used for each experimental point.Hemoglobin content quantification and histological analysis were performedas described previously (20).

The same assay was used to assess the anti-angiogenic activity of recom-binant TRAIL or inhibin-ba/activin-A (100 ng per implant) added to liquidmatrigel, using a vascular endothelial growth factor (VEGF) (50 ng/ml) þTNF-alpha (2 ng/ml) þ heparin (26 U/ml) cocktail as angiogenesis inducersand eight implants for each experimental point, as described previously (14).

In vivo tumor growth

Kaposi’s sarcoma tumor growth was obtained as described previously (14).LA (86 lg per mouse per day) was administered in sterilized drinking waterto nude mice, starting from the day of tumor cells injection. Tumor size wasassessed on days 9, 11, 13, 16, 18, 20, 23, 25 and 27. All animals (eight foreach experimental point) were killed when the first tumor reached the size of�1 cm3. Samples were paraffin embedded and stained with hematoxylin andeosin for histological analysis. Vessel quantification was obtained by analyzing10 lm-thick tissue slices under a fluorescence microscope (100� total magni-fication) with a triple-band filter: thick slices do not loose erythrocytes from ves-sels which, in turn, show a strong orange auto-fluorescence; CD-31 staining wasalso performed on some 3 lm-thin slices of tumor samples (with an anti-CD-31rat monoclonal antibody, clone BM4086B, Acris, Germany) for confirmation.

Growth assay

At day 0, cells were plated in 96-microwell plates at 1000 cells in 200 ll ofculture medium. LA was added to the cells at different concentrations (10, 50,250, 500 and 1000 lM). Due to the different growth rates, KS-Imm cells weremonitored at days 2 and 4, whereas HUVECs were quantified at days 3 and 5.Indirect optical density quantification was obtained fixing/staining the cells for20 min with a crystal violet solution as described previously (14).

Chemotaxis

Cell migration was assessed in Boyden chambers (Costar no longer produceschemotaxis chambers, Neuro Probe, Gaithersburg, MD 20877 USA) usingserum-free 3T3-conditioned medium (for KS-Imm) or KS-CM (for HUVECs)as chemoattractants as described previously (14,16). Cells were pre-treatedwith LA at the indicated doses and times, and treated with the same concen-tration of LA used for preconditioning also during the chemotaxis assay. Cellmigration was assessed by direct counting of five to eight unit fields per filter ordensitometric scanning of whole filter surface. Each test was performed intriplicate and repeated three times.

Adhesion

The 96-microwell plates for bacterial culture (Nunc GmbH, Wiesbaden, Germany)were pre-coated with 100 ll per well of water containing gelatin (50 lg/ml). After1 h, all coating solution was removed and HUVE or KS-Imm cells were plated(3000 cells/200 ml per well) in Dulbecco’s modified Eagle’s medium 1% bovineserum albumin. Cell cultures were treated with LA either only during the test orfor 24 h or 4 days before the test. Cells were incubated for 2 h at 37�C in 5% CO2

and eventually quantified by crystal violet solution as above.

Apoptosis

Cytoplasmic histone-associated DNA fragmentation was evaluated by the CellDeath Detection Kit, Roche, Milan, Italy according to manufacturer’s instruc-tions. KS-Imm and HUVE cells were plated (20 000 cells/ml per well) incomplete medium and allowed to grow for 48 h in 24-well plates. After thisstabilization period, the culture medium was changed and supplemented eitherwith LA (0, 10, 50, 250, 500 and 1000 lg/ml) or vincristine (0.5–1 lM).

Preparation of RNA and cRNA

Total RNA for microarray experiments was isolated from HUVECs treated for5 h with 200 lM LA as described previously (21). RNA samples were similarlyprepared for real-time polymerase chain reaction (PCR) testing using KS-Immand HUVE cells treated with 10, 50, 250, 500 and 1000 mM LA for 24 h, 48 hand 4 days.

Gene chip microarray analysis and data normalization

The labeled cRNAs were used for screening GeneChip Human GenomeU95Av2 arrays (Affymetrix, Santa Clara, CA). Data were collected using anAffymetrix scanner. The raw data of 32 features for each probe set wereanalyzed by Microarray Analysis Suite MAS 5.0. This included a statisticalanalysis of data consistency (one-sided Wilcoxon’s signed rank test) withina probe set that yielded a P value for the expression call (present, absent and

marginal) and the expression change between treated and control samples.Expression data and the expression and change P values were imported intothe GeneSpring 4.2 microarray data analysis program (SiliconGenetics, RedwoodCity, CA). For normalization, the 50th percentile of all measurements was usedas a positive control for each sample, each measurement for each gene wasdivided by this synthetic positive control, assuming that this was at least 10.The bottom 10th percentile was used as a test for correct background sub-traction. This was never less than the negative of the synthetic positive control.Each gene was normalized to itself by making a synthetic positive control forthat gene, and dividing all measurements for that gene by this positive control,assuming it was at least 0.01. The synthetic control was the median of thegene’s expression values over all the samples. Lastly, normalized values below0 were set to 0. The comparison of treated and mock-treated samples waslimited to genes that were expressed above a threshold level of 20 with a de-tection P value , 0.005 in both arrays in at least one condition. Genes thatshowed an at least 1.6-fold expression change in both arrays of the drug-treatedversus the mock-treated sample with a change P value , 0.05 were consideredas consistently changed. Gene lists for functional annotations were createdusing the NetAffx (Affymetrix) annotation tool.

Real-time PCR time-lapse

Total RNAs were isolated from control and LA-treated KS-Imm and HUVEcells, grown in complete medium, using the RNeasy Mini Kit (Qiagen, Milano,Italy). Each sample was amplified in triplicate. Reverse transcription wasperformed with oligo-dT primers and mRNA expression for selected geneswas analyzed by quantitative real-time reverse transcription–PCR by using thefollowing specific primers: interleukin 8—forward: gacaagagccaggaagaaacand reverse: gctcgtaggtcagaaagatgtg; TRAIL—forward: tcagagagtagcagctcacand reverse: ccttgatgattcccaggagt; PIM-2h—forward: ctcacagatcgactccaggtgand reverse: actttccatagcagtgcgactt; GPCR-kinase 5—forward: aactgggaga-gaaagggaagg and reverse: gttctttgcacggcttctgtag; thioredoxin reductase1—forward: atggaagaacatggcatcaagt and reverse: cctcactattggtggactgagc;thioredoxin-interacting protein—forward:taattggcagcagatcaggtc and reverse:acatccatatagcagggagga; heregulin-b2—forward: tcagtatccacagaaggagcaa andreverse: gtctttcaccatgaagcactcc; ephrin-B2—forward: cagacaagagccatgaagatcand reverse: caaagggacttgttgtcgaact; heme oxygenase-1—forward: tgatagaa-gaggccaagactgc and reverse: ggcagaatcttgcactttgttg; inhibin-ba/activin-A—for-ward: aacgggtatgtggagatagagga and reverse: aaatctcgaagtgcagcgtct.

cDNAs were amplified as described previously (21). Housekeeping geneRPII was used for data normalization, and relative expression values withstandard errors and statistical comparisons (unpaired two-tailed t-test) wereobtained using Qgene software (http://www.qgene.org).

Real-time PCR was also used to analyze the effect of 200 lM LA, on VEGF-A mRNA expression in KS-Imm treated for 24 h, 48 h or 4 days.

Fluorescence microscopy

HUVECs seeded on gelatin-coated multi-well chamber slides (LabTek, Nunc,Naperville, IL) were cultured at 37�C, pre-treated for 24 h, 48 h or 4 days withLA at the indicated doses. At the end of the incubation, mitochondria po-larization was stained for 30 min at 37�C with the fluorescent probeMitoTrackerRed CMX ROS (Molecular Probes, Invitrogen, Milan, Italy) di-luted in culture medium at 50 nM. Cells were then fixed with 4% paraformal-dehyde in phosphate-buffered saline (PBS) for 5 min, and counterstained with1 lg/ml 4#,6-diamidino-2-phenylindole (DAPI) (Sigma–Aldrich) in PBS.

For immunofluorescence detection of thioredoxin reductase, a monoclonalantibody (Lab-Frontier, Seoul, Korea) at 1:100 dilution in PBS–1% horseserum was incubated for 1 h with HUVE cells that had been treated withLA as above and fixed and permeabilized with cold methanol for 4 min at�20�C and blocked in PBS–10% horse serum for 10 min. Cells were washedthree times with PBS and incubated with fluoresceinated secondary anti-mouseantibody (Amersham-Pharmacia Biotech, Milano, Italy) at 1:200 dilution inPBS–1% horse serum for 30 min. Cells were then counterstained with DAPI(1 lg/ml) for 5 min and washed in PBS. The slides were mounted withProLong antifade reagent (Molecular Probes) and viewed in a Leica DM Lepifluorescence microscope at 40� or 100� magnification. Simple DAPIstaining was used to show the accumulation of the auto-fluorescent LA inHUVE cells.

Western blot analysis of proteins and cell signaling

HUVECs were cultured in standard conditions or treated with 250 lM LA, for24 h, 48 h or 4 days in complete medium without further medium/LA addition.Cell samples were prepared and blotted as described previously (15). Themembranes were then incubated with antibodies at the appropriate dilutionsin 5% powdered skim milk dissolved in 25 mM Tris-buffered saline (TBS)containing 0.15 M NaCl, 0.05% Tween-20, if not otherwise stated. The fol-lowing anti-human antibodies were used at the indicated dilutions: mousemonoclonal anti-inhibin-ba (Serotec Ltd, Oxford, UK) and mouse monoclonal

The anti-angiogenic activity of lipoic acid

1009


http://www.qgene.org

anti-thioredoxin reductase 1 (Lab-Frontier) 1:200 diluted in 5% skimmed milk,TBS–Tween 0.05%; rabbit anti-p-Akt and anti-Akt, rabbit anti-p-p38 and p-38MAPK, rabbit anti-p-Fak and Fak and rabbit anti-p44/42 MAPK (Cell Signal-ing, Beverly, MA) 1:1000 diluted in 5% bovine serum albumin, TBS–Tween0.1%; rabbit anti-p-Erk1/2 (Cell Signaling) 1:1000 diluted in 5% skimmed milk,TBS–Tween 0.1%; rabbit anti-GAPDH horseradish peroxidase conjugated,Novus, Littleton, CO) 1:2500 diluted in 5% skimmed milk, TBS–Tween 0.05%.The antibodies were reacted with the membranes for 1 h (anti-GAPDH), or 2 h(anti-inhibin), at room temperature, or overnight at 4�C (others). Secondary horse-radish peroxidase-labeled anti-rabbit or anti-mouse antibodies (Amersham-Pharmacia Biotech) were used at 1:5000 dilution in 5% skim milk powder,TBS–Tween 0.05%. The immune reaction was revealed by the ECL-Plus de-tection system (Amersham-Pharmacia Biotech). The test was repeated threetimes on three different HUVEC preparations (experimental replicates).

Results

In vivo angiogenesis

Since previous data indicated that LA inhibited TNF-alpha-induced IkB kinase activation and nuclear translocation of NF-kB(9), and that studies by our group (14,15,21) and others (22) indicatethat this is a common and key effect of many angioprevention agents,we studied the effect of LA on angiogenesis in vivo. KS-CM in amatrigel sponge was injected subcutaneously in c57/black mice (syn-genic to matrigel components). In this condition, a rapid angiogenicresponse was observed within 4 days: quantification of the angiogenicresponse by hemoglobin content detection demonstrated that LA-treated implants contained �1/12 of hemoglobin as compared withuntreated controls (Figure 1a), a significant reduction in hemoglobincontent was confirmed in the histology observations, where controlmatrigel þ KS-CM became filled with infiltrating cells, and largevascular lacunae lined by endothelium (Figure 1b). Whereas, whenmice were treated with LA (86 lg per mouse per day, correspondingto �150 mg per man per day, one-fourth of the standard 600 mgdosage used for standard therapy or body building), a dramatic blockof infiltration and angiogenesis was observed. The histology of

these samples shows an almost empty matrix with few cells scatteredin it (Figure 1c). Statistical significance of the angiogenic status foundin treated versus untreated mice was assessed by Student’s t-test(P 5 0.0016).

In vivo tumor growth

When KS-Imm cells were injected in nude mice, they formed large,non-metastatic, tumors, characterized by evident vascularization, sim-ilar to human KS lesions. The daily administration of LA (86 lg permouse per day, as in the angiogenesis assay) in drinking water causeda significant reduction in the KS tumor growth rate (Figure 2a). Thedifferences in tumor growth between LA-treated and -untreated con-trols were always statistically significant (P � 0.05, Student’s t-test) starting from the ninth day of the experiment, except for the timepoint at the sixteenth day (P 5 0.078). In our experience, this growthrepression would be consistent with an inhibitory effect linked to thereduced recruitment of endothelial cells seen in the matrigel plugassays. This hypothesis was reinforced by the histological analysisof collected KS-Imm tumors; in the presence of LA, vessels were lessfrequent and degenerating tumor cells with picnotic nuclei were oftenobserved (Figure 2b). Direct vessel quantification showed a 56% re-duction when controls were compared with LA-treated tumors (14.9 ±1.3 versus 8.3 ± 1.8 per field, respectively; P 5 0.0013, Student’s

Fig. 1. Effects of LA on angiogenesis in vivo. The injection of matrigelsponges, containing KS-CM, in mice causes a rapid angiogenic response thatis completely abolished by LA, administered in the drinking water. The graph(a) reports the mean hemoglobin content of recovered gels. The histologyshows poor cell recruitment inside the implants of mice supplemented withLA (c), whereas control samples are infiltrated with cells and filled withvessels and lacunae (b).

Fig. 2. Effects of LA on KS-Imm tumor growth in vivo. Mice injectedwith KS-Imm developed large, richly vascularized, primary tumors;when mice were treated with LA, in the drinking water, tumor onset wasdelayed and tumor growth was inhibited (a). Histological examinationshowed poor vascularization and large necrotic areas in LA-treatedtumors (b).

P.Larghero et al.

1010


t-test). It is important to notice that a putative anti-angiogenic agentable to block endothelial recruitment without any toxic effect onthe endothelium itself would be the best choice for any long-lastingtreatment of angiogenesis-sustained pathologies. To gain insights onthe effects of LA directly on endothelial and KS cells, we tested theeffects of this molecule in vitro.

Growth assays

LA was tested for its ability to modulate KS-Imm and HUVE cellsgrowth. Interestingly, LA showed a dual effect according to the dif-ferent concentrations used, the time and the cell type (Figure 3). OnKS-Imm cells, LA showed no inhibitory effect after 48 h of culture atany of the doses tested (10, 50, 250, 500 and 1000 lM), whereasa surprising growth-inducing activity was observed at 4 days withdoses ranging from 10 to 250 lM. The growth-promoting activitywas higher at the lower dose of LA tested (10 lM). High LA doses(500 and 1000 lM) showed, respectively, not significant or clear in-hibitory activity as compared with untreated controls. Student’s t-testwas used to weigh the differences between untreated and LA-treatedKS-Imm cells: the statistical significance was obtained both at 2 and 4days for almost all the experimental points (P values � 0.0149),except for 500 lM LA, at the fourth day (P 5 0.3919). In HUVEcells, LA showed significant modulation of cell growth, although withlonger incubation times than those observed with KS-Imm (Figure 3).In these cells, the growth-promoting activity of LA was weak, whereashigh doses (500–1000 lM) reduced HUVEC proliferation, but onlyafter 5 days of culture. In HUVECs, at the third day no statisticallysignificant variations were found, whereas at the fifth day, only 50 lMLA did not reach significance (P 5 0.119). For the other doses of LA,the P value was �0.019 (Student’s t-test).

Taken together, these observations show a null or positive effects oflow-dose LA on cell growth, whereas only high-dose LA can exertsome inhibitory potential after several days of treatment.

The same experimental setting was applied to other tumor cell lines ofdifferent histotypes (prostate: PC3; breast: MDA435; retina: Y79; my-eloid: K562) to test if the inhibiting/promoting activity of LA is linked toparticular cells or is a common feature of this molecule. Only one(MDA435) among these cell lines was affected by LA treatment, witha partial inhibition of growth at high doses, but no promotion was ob-served at low doses of LA (data not shown). According to these results,we can deduce that a therapeutic effect of LA against in vivo tumorgrowth was not mediated by cytotoxicity on the tumor cells; in contrast,it was able to repress KS-Imm tumor growth in vivo even though itstimulated growth of these cells in vitro.

Chemotaxis

Both tumor metastasis and angiogenesis imply cell migration, the firstto spread tumor cells out of primary site and the second to makeendothelial cell cross basement membrane and move towards an an-giogenic stimulus. When LA was tested treating KS-Imm for 24 h,only a weak inhibition of migration was observed using the highestdose (1000 lM). In this case, the migration was �65% of controls(Figure 4a). When KS-Imm cells were exposed to LA for 4 days,a clear dose–response curve was observed (Figure 4a), with LA actingas an inhibitor at all doses tested. Even in this experimental setting,the highest inhibitory activity was observed with 1000 lM LA.Student’s t-test was used to weigh the differences between untreatedand LA-treated KS-Imm cells: statistical significance at 24 h wasobtained only with 1000 lM LA (P 5 0.04), and at 4 days, all thedoses of LA gained significance (P values � 0.04).

When endothelial cells were treated for 24 h, 48 h or 4 days withLA, a different response was observed (Figure 4b). At 24 h, LA causeda dose-dependent inhibition of cell migration. After 48 h treatment,LA was apparently less able to inhibit HUVEC migration at the lower,non-cytotoxic, doses; in fact, only 500 and 1000 lM LA, inducesignificant reductions of cell chemotaxis. After 4 days of LA treat-ment, HUVEC response is even more altered: 10 lM LA induceda significant increase in the chemotactic ability of HUVECs, andthe same increasing trend—though not statistically significant—wasobserved for 50 lM LA. Higher doses of LA (250–1000 lM) wereclear inhibitors of HUVEC migration, the higher dose leading to thecomplete inhibition of cell response. In HUVECs, statistical signifi-cance at 24 h was obtained with 50, 500 and 1000 lM LA (P values �0.041), at 48 h with 500 and 1000 lM LA (P values � 0.029), and at 4days with 10, 250, 500 and 1000 lM LA (P values � 0.010).

These data suggest that LA exerts a short-term dose-dependent -inhibitory effect on endothelial cell migration, whereas a long-termtreatment can cause opposite effects according to the dose used. Theactivity on KS-Imm cell migration seems almost subverted, with noinhibition by short incubation and a dose-dependent inhibition after 4days of LA treatment.

Adhesion

An altered adhesion to the substrate could explain the reduced migra-tion of LA-treated cells. To check this hypothesis, we tested bothHUVE and KS-Imm cells for their ability to adhere to a gelatin sub-strate in the presence or absence of LA. No positive or negative effectwas observed at any dose of LA tested on both cell lines if LA wasused during the test without cell pretreatment (data not shown). This

Fig. 3. Effects of LA on cell growth. KS-Imm and HUVECs were treated with LA in complete medium; low-dose LA stimulate cell growth as compared withcontrols, whereas high-dose LA shows inhibitory properties, more evident in HUVECs.


1011


observation excludes a possible physical, direct, interference playedby LA on adhesion molecules or matrix coating.

When KS-Imm were treated for 24 h with LA at 10, 250 and 1000lM, their adhesive properties on gelatin appeared unaffected (Figure5a). The same concentrations of LA acted as dose-dependent inhibitorsafter 4 days of treatment. Student’s t-test was used to weigh the differ-ences between untreated and LA-treated KS-Imm cells: no statisticalsignificance was obtained at 24 h, and at 4 days, all LA treatmentsproduced significant data (P values � 0.012). The effect of LA onHUVEC adhesion after 24 h of treatment was weak, with a slightincrease of adhesion with 10 and 250 lM LA, and a decrease with1000 lM LA. After 4 days of treatment with 10 lM LA, the adhesionon gelatin was further increased (132%), whereas the 1000 lM dosealmost completely abolished cell adhesion (Figure 5b). Statistical sig-nificance was obtained already at 24 h for all LA treatments (P values� 0.041), and at 4 days for 10 and 1000 lM LA (P values � 0.004).

These observations suggest a possible contribution of adhesivepathways or cytoskeleton assembly to the inhibitory effects of LAon HUVECs, though the apparently opposite effects observed onHUVECs with 10 lM LA, in adhesion and chemotaxis experimentsat 24 h and 4 days suggest a non-univocal linkage between adhesive–chemotactic properties and LA inhibition.

Apoptosis

The growth assays showed that LA, depending on the doses tested,can either promote cell growth or act as a cytostatic agent. Accord-ingly, we examined the effects of LA on apoptosis. LA used aloneon KS-Imm at the dose of 250 or 1000 lM for 24 h did not modulateapoptosis, whereas 4 days of treatment induced a statistically sig-nificant increase of the apoptotic rate with 1000 lM LA (Figure 6a).When KS-Imm cells were treated with 0.5 lM vincristine, a strongapoptotic rate was observed. LA did not show any protective effectat 250 and 1000 lM, when the cells were treated 24 h or 4 daysbefore vincristine challenge (Figure 6b). Student’s t-test was usedto weigh the differences between untreated and LA-treatedKS-Imm cells: no statistical significance was obtained at 24 h, and at4 days, 1000 lM LA produced a significant variation (P value 50.010). No significant variation was found in KS-Imm samples treatedwith vincristine.

HUVE cells were almost unaffected after 24 h of treatment with10, 50, 250, 500 or 1000 lM LA (Figure 6c), with a minor, butstatistically relevant, reduction of the basal apoptotic rate with LAdoses at or over 50 lM. When HUVECs were treated with 1 lMvincristine in complete medium, a slight increase of apoptosis wasobserved (1.2-fold as compared with untreated controls). Again, LAacted as a suppressor of apoptosis with doses over 50 lM. In HU-VECs, statistical significance was achieved with LA doses �50 lM,both with (P values � 0.045) or without (P values � 0.036) vincristinetreatment.

Fig. 4. Chemotaxis of KS-Imm and HUVE cells. (a) A 24 h exposition of KS-Imm to increasing doses of LA causes little inhibitory effects, statisticallysignificant only at the higher dose (1000 lM), whereas after 4 days,a significant dose-dependent inhibition is observed. (b) HUVECs treated withincreasing doses of LA show different cell responses according to the durationof treatment. After 24 h, a dose–response inhibition curve is observed(statistically significant starting from the 50 lM dose), whereas at 4 days,low-dose LA causes a little increase of cell migration (differences in migrationare statistically significant for the 10, 50, 250, 500 and 1000 lM doses).

Fig. 5. Effects of LA on cell adhesion. Short-term treatment of KS-Imm(a) with increasing doses of LA does not influence cell adhesion to gelatin,whereas after 4 days, a statistically significant, dose-dependent inhibition isobserved. In HUVECs, LA acts similarly (b), but is able to induce a slightincrease of cell adhesion at the lower dose.

P.Larghero et al.

1012


Gene chip microarray analysis

Although LA appeared to be an efficacious anti-angiogenic agentin vivo and to inhibit endothelial cell migration in vitro, the contrast-

ing effects observed with diverse doses in vitro led us to further in-vestigate the mechanisms underlying these effects. Microarray-basedanalysis of early genes, expressed in HUVE cells treated for 5 h withLA and showing at least 1.6-fold expression change during LA (200lM/5 h) treatment, produced a list of 51 genes whose expression wassignificantly altered, where 30 appeared to be up-regulated by LA and21 down-regulated. Using a 2-fold increase or decrease threshold, thenumber of modulated genes counted only nine and six records, re-spectively, indicating that LA, at low dose for short periods, has nodrastic effects on gene expression in endothelial cells. This would beconsistent with limited toxicity of LA.

Most of the genes modulated by LA can be categorized as beingimplicated in cell cycle/proliferation/apoptosis or oxidative stress/in-flammation. These two groups account at least for 61% (31 of 51) ofthe genes affected by LA. As some of the genes regulated by LA havean unknown function, this percentage could be underestimated.

Among the 18 genes involved in cell cycle/proliferation/apoptosis,we found up-regulated proliferation inducers like heregulin-b2/neuregulin, PIM-2h, Ki-67 and neuropilin, accompanied by modula-tors/effectors of cell differentiation such as estrogen-responsive Bbox protein, inhibin-ba/activin-A, protein kinase C-alpha and neuralcell adhesion molecule (Table I). On the other hand, cell cycle inhib-itors and apoptosis inducers including Fas/APO-1, Rb-1, G protein-coupled receptor (GPCR)-kinase 5, CCR4-NOT, ephrin-B2 andTRAIL/APO-2L appeared to be down-regulated (Table I). This generegulation �profile� would suggest a commitment of short-term LA-treated endothelial cell towards a healthy, differentiating/proliferatingphenotype.

We attributed 13 genes to the oxidative stress/inflammation list(Table II), most of these RNAs were up-regulated. Heme oxygenase-1 (HO-1), protecting from oxidative damage, appears as the mostintensely regulated gene by LA, with a 6.4-fold increase, followedby thioredoxin reductase 1 (2.3�), whose gene product is involved inLA reduction to the active form dihydrolipoic acid. The modulationof genes in this group would apparently indicate a cellular responseinduced by inflammation/oxidative stress with production of protec-tive proteins (HO-1, C/Ebp-homologous protein, nicotinamide aden-osine dinucleotide phosphate (NADPH) dehydrogenase, humanmineralocorticoid receptor) and inflammatory cytokines/receptors (IL-8 and IL-6 signal transducer gp130). Interestingly, unlike that previouslyobserved for NAC and to a lesser extent EGCG (21), we did not observeobvious modulation of genes related to the NF-kB pathway.

Fig. 6. Evaluation of inducing/protective effects of LA on apoptosis. Onlya long-term treatment with high-dose LA can increase the apoptotic index ofKS-Imm cells (a), but LA does not show protective effects againstvincristine-induced apoptosis. After a 24 h treatment with LA, HUVECs(b) show a slight decrease of apoptotic rate, and in these cells LA partiallyreduces vincristine-induced apoptotic rate.

Table I. Microarray data: list of LA-regulated genes linked to cell cycle/proliferation/apoptosis, in HUVECs

Name Expression levelLA-treated/control

Possible role in angiogenesis

Heregulin-b2/neuregulin 2.5� ErbB3/4 ligand, pro-angiogenic (autocrine).PIM-2 proto-oncogene homolog (PIM-2h) 2.4� Ser/Thr protein kinase, negative regulator of STAT6. Induces proliferation.Estrogen-responsive B box protein 1.9� Enhances early differentiation process, do not affect proliferation.Inhibin-ba/activin-A 1.8� Anti-angiogenic, tumor growth factor-Fb family, (autocrine).Protein kinase C-alpha 1.7� Induced by . intracellular [Caþþ]. Increases endothelial cell migration,

mediates epithelium differentiation.Ki-67 nuclear antigen 1.7� DNA-binding, expressed in proliferating cells.Neuropilin-like protein 1.6� VEGF co-receptor, pro-angiogenic.Neuronal cell adhesion molecule 1.6� Involved in 3D vessel organization.TNFr superfamily, member 6/Fas/APO-1 0.6� Apoptosis inducer receptor.Retinoblastoma 1 0.6� Cell cycle control.GPCR-kinase 5 0.6� Blocks the signal mediated by GPCR–GPCR ligand interaction.LIM domain only 2 0.6� Cell differentiation.ATP-synthase 0.6� Angiostatin receptor.MADS box transcription enhancer factor 2 0.6� Modulate organization of endothelial and smooth muscle cells into vascular

plexus.CCR4-NOT transcription complex, subunit 8 0.5� Negative control of cell proliferation.Activation-induced C-type lectin 0.5� Lectin, activation marker for leucocytes.Ephrin-B2 0.5� Histological marker of neoangiogenesis, in vitro blocks HUVE cells

response to VEGF and Ang1.TNF-ligand superfamily, member 10/TRAIL/APO-2L 0.3� Apoptosis inducer.


1013


LA-induced gene expression quantified by real-time PCR

We selected 10 representative genes among those found modulated inthe microarray analysis and used them to confirm the microarray dataand execute more detailed analyses of the effects of LA over time oncells in vitro. The genes chosen were as follows—for cell cycle/pro-liferation/apoptosis: TRAIL, heregulin-b2, PIM-2h, inhibin-ba/activin-A, ephrin-B2 and GPCR-kinase 5; for oxidative stress/inflammation:thioredoxin reductase 1, thioredoxin-interacting protein, HO-1 and IL-8. A short-term exposure to LA does not seem to strongly influence RNAexpression in microarray analysis, and in vitro tests described above onKS-Imm and HUVECs show significant modulatory activity by LA

only in a time- and dose-dependent mode. We consequently used LAat 200 and 1000 lM concentrations and treatment times of 24 h, 48 hand 4 days. Statistical analysis of each time point compared with thebasal level of gene expression was performed by Student’s t-test, andthe complete list of P values is shown in Table III.

Cell cycle/proliferation/apoptosis

TRAIL. This gene was found weakly down-regulated in KS-Immtreated with 200 lM LA for 24 h, and a longer treatment caused thisgene to revert to basal levels of expression (Figure 7); on the contrary,1000 lM LA induced a time-dependent increase of TRAIL mRNA

Table II. Microarray data: list of LA-regulated genes linked to oxidative stress/inflammation, in HUVECs

Name Expression levelLA-treated/control

Possible role in angiogenesis

HO-1 6.4� Protects from oxidative damage, binds NO.Thioredoxin reductase 1 2.3� Regulates ROS-mediated signaling, reduces LA to Dihydrolipoic acid DHLA.Homocysteine-inducible, endoplasmic reticulumstress-inducible, ubiquitin-like domain member 1

2.2� Stress induced.

Solute carrier family 7 member 11 2.1� Cystine/glutamate antiporter, increases intracellular glutathione.C-EBP homologous protein (CHOP) 2.1� Stress response to endoplasmic reticulum damage, induced by bFGF in

endothelium.UDP-glucose dehydrogenase 1.9� Proteoglycan synthesis, induced by inflammatory cytokines.Interleukin 8 1.7� Inflammatory chemokine, recruits neutrophils, angiogenic.NADPH dehydrogenase/quinone 1 1.7� Detoxification response, gene regulated by an antioxidant response element.IL-6 signal transducer (gp130, oncostatin M receptor) 1.7� Inflammation.Glutamate-cysteine ligase, modifier subunit 1.6� Regulates glutathione synthesis.Human mineralocorticoid receptor 1.6� Prevents up-regulation of vascular ET-1, restores NO-mediated endothelial

dysfunction.NADPH-dependent isocitrate dehydrogenase 0.6� NADPH production, glutathione synthesis.Thioredoxin-interacting protein 0.6� Binds/inhibits thioredoxin.

Table III. Statistical validation of real-time PCR analysis

Gene Student’s t-test P values

KS-Imm,200 lM LA

KS-Imm,1 mM LA

HUVEC,200 lM LA

HUVEC,1 mM LA

TRAIL 24 h versus control 0.0157 0.0464 0.0010 0.021048 h versus control 0.2142 0.0178 0.0005 0.01964 days versus control 0.1356 0.0066 0.0276 0.2693

Heregulin b 2 24 h versus control 0.0001 0.8708 0.4000 0.264048 h versus control 0.0065 0.0304 0.4715 0.00394 days versus control 0.0007 0.9323 0.4827 0.0169

PIM-2 homolog 24 h versus control 0.1790 0.0997 0.3498 0.495448 h versus control 0.8525 0.0474 0.2212 0.00584 days versus control 0.4821 0.5706 0.0240 0.0017

Inhibin-ba 24 h versus control 0.0250 0.3508 0.0001 0.025548 h versus control 0.0466 0.0147 0.0005 0.00994 days versus control 0.0129 0.0010 0.0001 0.0033

Ephrin B 2 24 h versus control 0.0281 0.4045 0.0064 0.016148 h versus control 0.0330 0.7256 0.0017 0.00174 days versus control 0.1180 0.0349 0.0049 0.0052

Thioredoxin reductase 1 24 h versus control 0.0018 0.0005 0.0004 0.0002548 h versus control 0.0106 0.0146 0.0024 0.001624 days versus control 0.0108 0.0018 0.0031 5 � 10-9

Thioredoxin-interacting protein 24 h versus control 0.0001 0.3050 0.0188 0.243948 h versus control 0.0000 0.0231 0.0331 0.00014 days versus control 0.0003 0.3208 0.0572 0.0001

Interleukin 8 24 h versus control 0.0042 0.9037 0.0502 0.000048 h versus control 0.0039 0.0122 0.0001 0.00544 days versus control 0.0034 0.0060 0.0128 0.0000

Heme oxygenase-1 24 h versus control 0.9577 0.3225 0.0050 0.224848 h versus control 0.3698 0.0349 0.0000 0.01994 days versus control 0.3936 0.7349 0.0059 0.0266

GPCR-kinase 5 24 h versus control 0.9442 0.8445 0.3110 0.061948 h versus control 0.4735 0.0045 0.0366 0.09224 days versus control 0.0177 0.0245 0.0007 0.0001

P.Larghero et al.

1014


reaching 9.85-fold levels as compared with controls after 4 days. InHUVECs, TRAIL was always down-regulated at 24 h and, as observedfor KS-Imm, a longer treatment caused this gene to revert to basallevels, though this modulation was similar for both LA doses.

Heregulin b 2. In KS-Imm, this gene was weakly up-regulated onlywith 200 lM LA (2.35 fold at 24 h), maintaining similar expressionlevels in the following days. High-dose LA did not affect the expres-sion of this gene. In HUVECs, heregulin b 2 was not modulatedconsistently in the long term, despite the up-regulation observed inthe microarray screening at 5 h treatment with 200 lM LA.

PIM-2h. Although KS-Imm did not show any modulation of thismRNA, in HUVE cells, PIM-2h was significantly up-regulated after48 h (maximum value 3-fold at 4 days with 1000 lM LA).

Inhibin-ba/activin-A. In KS-Imm, both 200 and 1000 lM LA wereable to induce a significant increase of this mRNA after 48 h, with high-dose LA being more active (8.62-fold at 4 days). In HUVECs, 200 lMLA induced the down-regulation of this mRNA after 24 h that wasmaintained at lower levels (0.28-, 0.29- and 0.26-folds at 24 h, 48 hand 4 days). On the contrary, high-dose LA caused a partial up-regu-lation of inhibin-ba/activin-A, with a 2.7-fold increase after 4 days ofLA treatment.

Ephrin B 2. This gene was poorly affected in KS-Imm, whereas it wasdown-regulated in HUVECs treated with 200 lM LA (0.14-fold at 4days). High-dose LA was able to cause a weaker down-regulation(0.27-fold at 4 days).

GPCR-kinase 5. This gene was unaffected in KS-Imm, whereas itreached significant modulation in HUVECs, where it appeared to beconsistently up-regulated after 4 days (2.77- and 8.00-fold for 200 and1000 lM LA, respectively).

Oxidative stress/inflammation

The first interesting observation was obtained by observing the ex-pression levels of thioredoxin reductase, the enzyme directly actingon LA (reducing it to the dihydrolipoic acid (DHLA) form). This genewas not only up-regulated upon LA treatment but also represented themost intensely expressed gene in both KS-Imm and HUVECs amongthe 10 selected genes (the mean normalized expression of thioredoxinreductase 1 in untreated HUVECs and KS-Imm was 2 to 3 log higheras compared with the levels of the other genes, data not shown).

Thioredoxin reductase 1. This gene was always up-regulated both inKS-Imm and HUVECs upon LA treatment (Figure 7), and the effectof LA appeared dose and time dependent indicating the strong corre-lation between LA and the enzyme used for its conversion to the

Fig. 7. Time and LA dose-dependent regulation of targeted genes, real-time PCR. Ten selected genes were studied for their expression levels at different times(24 h, 48 h and 4 days) and under different LA regimens (200 or 1000 lM) in KS-Imm and HUVECs. See Results for an extended description.


1015


reduced form, DHLA. Maximal modulating values ranged from 4.47-fold in KS-Imm (at 48 h) to 18.02-fold for HUVECs (at 4 days).

Thioredoxin-interacting protein. Although a short exposition to 200lM LA down-regulated this enzyme in the microarray analyses, after48 h of high-dose LA, this mRNA was strongly up-regulated in HU-VECs as a probable attempt to preserve NADPH reserves from com-plete consumption through LA reduction to DHLA by thioredoxinreductase 1. The response to 200 lM LA was weaker, whereas inKS-Imm, this dose of LA sustained a strong down-regulation of thisgene.

IL-8. In KS-Imm, IL-8 was time dependently up-regulated by LA, the200 lM dose being more effective (17.22-fold at 4 days). In HUVECs,the up-regulation was less powerful and more rapid when using high-dose LA (maximum 5.48-fold at 4 days).

Heme oxygenase 1. Unaffected in KS-Imm cells, this gene was up-regulated in HUVECs, with higher and more rapid modulation at 200lM LA (7.64-fold at 24 h).

These data show that LA activity is directly linked to the dose andtime of cell exposition, possibly causing opposite effects on cell be-havior. These data led us to further investigate specific aspects of LAeffects in cells in vitro.

VEGF-A expression in KS-Imm cells treated with LA

In endothelial cells, thioredoxin reductase 1 acts as an inducer of HO-1(23) and this molecule was described to induce VEGF expression(24,25); therefore, we decided to verify if this induction could alsobe active in KS-Imm. In vitro, control KS-Imm samples showed a com-parable expression of VEGF-A at 24 and 48 h of culture (mean nor-malized expressions were 1.15-02 and 1.13-02, respectively), whereasa doubling was observed after 4 days of culture (mean normalizedexpression was 2.28-02). In the presence of 200 lM LA, the mRNAof VEGF-A showed a weak and not statistically significant time-de-pendent increase, reaching—at 4 days—levels almost identical to un-treated controls (mean normalized expressions at 24 h, 48 h and 4 dayswere, 1.63-02, 1.97-02 and 2.32-02, respectively), suggesting that LA inKS-Imm does not act as a significant inducer of VEGF. This observa-tion can be linked to the lack of HO-1 induction described, for KS-Imm, in the previous paragraph.

Fluorescence microscopy

LA has already been described to be able to induce mitochondrialpermeabilization and oxygen consumption both in its oxidized andreduced (DHLA) form; this activity was accompanied by the produc-tion of reactive oxygen species able to induce mitochondrial damageand selective tumor cell death (26,27). These observations would beconsistent with the possible induction of genes linked to oxidative stressin HUVECs. To verify if LA is able to trigger mitochondrial activation,we treated HUVECs with different doses of LA, imaging at differenttimes the effective mitochondrial uptake of cationic, lipophilic fluoro-chrome chloromethyl-X–rosamine. LA (tested at 10, 250 and 1000 lM)elicited a similar activation of mitochondria, persistent for days (Figure8, 48 h treatment). At 1000 lM, LA partially accumulated inside thecells, forming insoluble auto-fluorescent granular aggregates (data notshown), probably participating to the toxic effects of the molecule seenafter 4 days. At this time point, the chloromethyl-X–rosamine uptakewas reduced.

Thioredoxin reductase gene showed high levels of expression andmodulation in HUVECs, and most likely represents a key mediator ofLA activity and toxicity as it is involved in LA/DHLA conversion. Westained HUVE cells with fluorescein isothiocyanate–anti-thioredoxinreductase 1 monoclonal antibody to verify modulation of thioredoxinreductase at the protein level. The expression of thioredoxin reductasewas found up-regulated in dose- and time-dependent manner upon LAtreatment (Figure 9).

Western blot analysis of proteins and cell signaling

The modulation of thioredoxin reductase 1 protein expression in HU-VECs was confirmed by western blotting, showing its time-dependentup-regulation upon LA (250 lM) treatment (Figure 10). Inhibin-ba/activin-A protein expression was also found up-regulated in a time-dependent fashion.

The up-regulation of mitochondrial activity together with the ex-pression of well-known markers of oxidative stress is suggestive of analtered endothelial phenotype. In vitro data have shown that LA exertsmodulatory effects on HUVE cell behavior; accordingly, we haveverified if central transduction pathways were affected by LA.

To reproduce the experimental conditions of the growth assays, weprepared both untreated and LA-treated HUVEC samples at 24 h, 48 hand 4 days, in complete medium analyzing the phosphorylation ofFAK, Erk1/2, Akt and p38-MAPK as compared with their amount oftotal protein and GAPDH expression (Figure 11). The western blotshows that phosphorylation of FAK is apparently down-regulated at24 h and 4 days, whereas it is similar to untreated cells at 48 h. Thephosphorylation status shown by Erk1/2 is weak as compared withuntreated controls and, at 48 h, little p-Erk is detectable. Akt showsa generally weak phosphorylation signal at 24 and 48 h and does notappear to be substantially modulated by LA. Phospho-P38 is detectableonly at the fourth day and it is strongly induced by LA treatment.

These biochemical data are in agreement with the in vitro observa-tions, showing that LA influences key transduction pathways involvedin cell growth, migration and adhesion.

Anti-angiogenic effects of TRAIL and inhibin-ba/activin-A in vivo

Among the potential anti-angiogenic molecules induced by LA in KS-Imm, we find inhibin-ba/activin-A and TRAIL; accordingly, we de-cided to test these proteins in the matrigel sponge model to evaluatetheir possible contribution to the anti-angiogenic effects of LA ob-served in vivo.

Both molecules showed a significant anti-angiogenic potential ex-plaining why LA can limit KS-Imm tumor growth in vivo at doses notexerting a direct effect on the tumor cell itself (Figure 12).

Discussion

Anti-angiogenic therapy through VEGF inhibition has significantlyincreased lifespan for several tumors in clinical trials when combinedto chemotherapy. These data suggest that further improvement in thisapproach has the potential to substantially extend patient lifespan. A

Fig. 8. Activation of mitochondria by increasing doses of LA, after 48 hexposition. HUVECs showed a strong increase of polarized mitochondria(showed by MitoTrackerRed internalization) at any of the LA doses tested.Similar results were obtained at 24 h (data not shown).

P.Larghero et al.

1016


gap in the phase I–II trials of anti-angiogenic molecules is linked tothe recruitment of patients with advanced malignant tumors, wherevascularization is at an advanced stage. How anti-VEGF therapy hasattained these results is a matter of discussion. Tumor endothelial cellsfrequently skip final differentiation steps, showing poor pericyte cov-ering and VEGF dependence (28,29); however, it is possible thattumor vessels under anti-VEGF therapy become �normalized� and de-liver chemotherapy more effectively (30). An alternative targeting oftumor vessels could be obtained with a chemopreventive setting, whereanti-angiogenic molecules—without side effects—could be admin-istered to patients with low-stage/-grade tumors, or after surgicaleradication or simply to people with increased risk of developingcancer. We are testing several �angiopreventive� drugs selected amongthe antioxidants: these molecules have high tolerability and theirchronic use in humans often already approved or are common di-etary practice.

We have shown previously that NAC and EGCG (a main constitu-ent of green tea extract) exert a true anti-invasive activity inhibitingtumor and endothelial cell invasion by matrix metalloproteinase-2neutralization (11,14,21,31,32). In addition, EGCG at high doses isalso able to block the cell cycle (33). Several of these antioxidant

angioprevention agents have the common effects of blocking theNF-kB and Akt pathways (11,14,21,31,32); since LA has been sug-gested to inhibit the NF-kB pathway (9), here we tested LA for pos-sible angioprevention activity. Our data show that LA can exertdifferent stimulatory/inhibitory activities on tumor and endothelialcells according to the dose and persistence of administration.

Fig. 9. Thioredoxin reductase 1 immunofluorescent staining in HUVECs. TrxR1 was strongly induced, in a time- and dose-dependent manner, in HUVECs treatedwith LA.

Fig. 10. Expression levels of inhibin-ba/activin-A and thioredoxinreductase 1 proteins in HUVECs treated with 250 lM LA: both proteinsare up-regulated in a time-dependent manner.

Fig. 11. Phosphorylation of key signaling molecules upon LA treatment inHUVE cells. Untreated (C) and LA-treated cells (LA), grown in completeculture medium for the indicated times, show variable degrees of responses.In particular, down-regulation of p-FAK and p-ERK1/2 is observed uponLA treatment, p-Akt is little affected, whereas phospho-P38 shows a strongup-regulation in long-term culture.


1017


An ideal antioxidant should be efficiently adsorbed from diet, con-verted—if necessary—into a usable form, helpful to establish/restoreother antioxidants and to be able to work in both lipidic and aqueousphases. In addition, it also needs to exert low toxicity. LA fulfills allthese requirements (34). It is not yet clear if LA is synthesized inmammals; whereas one study found LA synthase in mouse cells (35),in humans it is usually assumed with food (particularly red meat, liver,spinach and yeast). Our body does not need large amounts of LAcoenzyme; on the contrary, a constant additional supply is neededfor antioxidant or therapeutic activity. However, very little is knownabout the consequences of a long-term high-dose LA regimens andnothing is known about the activity of LA on most tumors.

Unlike other antioxidants, low, non-toxic doses of LA did not showany growth inhibitory effect on endothelial or tumor cells. This resultis supported by the observation of the almost unaffected phosphory-lation of Akt in LA-treated HUVECs, the lack of apoptosis inductionand the partial protective effects LA exerts on endothelial cell survivalin the presence of vincristine. In contrast to LA, angiopreventiveantioxidants, such as EGCG and Xanthohumol, inhibited Akt phos-phorylation and activation (14,15). In agreement with our data, Kowl-uru and Odenbach (36) have recently shown in a rat diabeticretinopathy model that LA administration inhibited capillary endothe-lial cell apoptosis in the retina. Further, Marsh et al. (37) observed thatLA administered with vitamin E can induce bcl-2 in endothelial cells,though this effect was not evident with LA alone. In a different cellmodel, Byun et al. (38) showed that LA inhibits TNF-alpha-inducedapoptosis in bone marrow stromal cells blocking c-jun N-terminalkinase and NF-kappaB. These data suggest that LA has a protectiveeffect on endothelial and other cell types.

On the other hand, we observed a cytostatic effect at the higherdoses of LA tested (0.5–1 mM); this effect was increased with time ofincubation, and after 4 days of treatment at 1 mM, most endothelialcells were damaged, showing a cytoplasmatic auto-fluorescent gran-ulation due to the accumulation of LA precipitates. When we testedtumor cells of different origins, the low-dose growth-promoting andhigh-dose inhibitory effect of LA was evident only in KS-Imm cells.

The null or promoting activity of LA on tumor cells is apparently incontrast with the results of van de Mark et al. (39) who reporteda tumor-specific apoptosis-inducing activity for LA on FaDu andJurkat tumor cell lines as well as on Ki-v-Ras-transformed Balb/c-3T3 as compared with normal cells. Our data show that diverse celltype can exhibit different levels of toleration or response to LA ac-cording to the dose and schedule of exposure, indicating that nogeneral rule can be made.

The prevalent inhibitory effect of LA on HUVE cell migrationcould be linked to the reduced phosphorylation of Fak, though this

pathway cannot completely clarify why long-term exposition to low-dose LA can cause a slight increase of cell motility.

The microarray-based screening suggested a selective clustering ofcellular responses to low-dose LA in the oxidative stress/inflamma-tion and growth/survival pathways. We selected a small group ofgenes to investigate further. KS-Imm and HUVECs showed substan-tial differences in the regulation of these mRNAs upon LA exposition:using an arbitrary cut-off of a ±3-fold gene modulation, in KS-Imm,PIM-2h, ephrin-B2, G-protein coupled receptor and HO-1 were notaffected. Low-dose LA in KS-Imm was able to influence only IL-8(strong up-regulation) and TrxIP (down-regulation), whereas high-dose LA up-regulated TRAIL, inhibin-ba/activin-A, TrxR1 and IL-8. These findings are consistent with the lack of inhibitory activity oflow-dose LA on KS-Imm in vitro, whereas high-dose LA for 4 daysinduces KS-Imm apoptosis and blocks the adhesive and chemotacticresponses of these cells. Several published data indicate that stress-related mediators like HO-1 and TrxR1 can act as tumor promoters,also inducing angiogenic responses (24,25,40–42); indeed, in endo-thelial cells themselves, this process is apparently active with complexregulations as both VEGF and TrxR1 are able to induce HO-1, whichin turn can switch on several pro-angiogenic factors (23,43–45).Several stress stimuli are able to induce HO-1 in endothelium, amongwhich these in vitro infection by Kaposi’s sarcoma-associated herpes-virus 8 (KSHV/HHV-8) (46); this activation is apparently mediated bythe HHV-8 gene product GPCR, codifying for a constitutively activechemokine-like receptor (47). Of course, the induction of HO-1 byLA constitutes a possible bias for its translation to clinic, though, inour hands, LA fails to activate HO-1 transcription in KS cells andVEGF production is consequently unaffected. In addition, it is impor-tant to note that, although in vitro endothelial cells are permanentlyexposed to LA, in vivo, this drug is rapidly cleared from circulation, sothe endothelial cell is exposed to it for very short periods reducing thestress activity.

Inhibin-ba/activin-A is released early in the cascade of circulatorycytokines during systemic inflammatory episodes, generally coinci-dent with TNF-alpha (48). Its up-regulation, along with IL-8, ina long-term high-dose LA regimen suggests a pro-inflammatory ac-tivity of LA. These data are in contrast to the inhibition of the NF-kBpathway reported previously (9). Inhibin-ba/activin-A has beenshown to inhibit both endothelial cell proliferation in vitro and angio-genesis in vivo (49), observations we confirmed by in vivo analyses,thus LA could push KS-Imm towards an anti-angiogenic phenotype.

LA mimics an oxidative signal in HUVECs even at subtoxic doses,triggering a form of cell resistance based on the down-regulation ofanti-angiogenic and pro-apoptotic mediators, and the up-regulation ofstress-induced proteins and phosphorylation of stress-related kinases[i.e. p38 (50)], induced by oxidative injury. These in vitro observationsare in agreement with the genomic/protein data as the inhibitoryeffects of LA on KS-Imm are observed only after a long-term/high-dose treatment, whereas low-dose LA is already active on HUVECsafter 24 h.

According to our observations, LA is able to limit KS tumor growthin vivo even without directly reducing KS cell survival/proliferation.This inhibition is probably, but not univocally, due to an anti-angio-genic mechanism as LA inhibits endothelial cell migration and indu-ces the expression of the anti-angiogenic factor inhibin-ba/activin-Aboth in HUVECs and in KS-Imm. In addition, LA-induced expressionof TRAIL in KS-Imm could act as an additional anti-angiogenicfactor as shown by the anti-angiogenic effect of TRAIL in vivo.

At low doses, LA did not effect endothelial cell survival or growth,suggesting that it is a safe drug for a long-term treatment of cancerpatients. The pro-/anti-antioxidant activities of LA, observed in ourtests and in other studies (51–54), are usually linked to the differentconcentrations of the drug tested, suggesting the necessity of a morepointed choice of the therapeutic regimen according to pathology, pos-sibly excluding the actual abuse of self-prescribed LA as a high-dosedietary supplement.

Taken together, our data show that LA represents a promising an-giopreventive drug whose action appears to be quite different from that

Fig. 12. Anti-angiogenic effects of inhibin-ba/activin-A and TRAIL onin vivo angiogenesis. The angiogenic response caused by the injection ofmatrigel additioned with angiogenic growth factors was inhibited by theenrichment with both inhibin-ba/activin-A and TRAIL, the latter being moreactive.

P.Larghero et al.

1018


of other antioxidant angiopreventive molecules. The current applica-tion for diabetes-associated neuropathies could, in the near future, beextended to cancer and other angiogenesis-sustained pathologies. Fur-ther studies are needed to identify the plethora of possible cellulartargets of this molecule, exclude those malignancies where LA couldact as a tumor promoter (i.e. inducing HO-1 and/or TrxR1 transcrip-tion) and distinguish other anti-angiogenic mechanisms it might trig-ger, including inflammation itself.

Acknowledgements

We wish to acknowledge Anna Buffa, Sebastiano Carlone and Luca Anfossofor helpful starting experiments, and Dr Anna Maria Colacci for helpful dis-cussion and suggestions. This study has been supported by grants from theCompagnia di San Paolo, the AIRC (Associazione Italiana per la Ricerca sulCancro), the MIUR Progetto FIRB and Ricerca Finalizzata, the PNR-Oncologiaand the Comitato Interministeriale per la Programmazione Economica (CIPE).

Conflict of Interest Statement: None declared.

References

1.Bustamante,J. et al. (1998) Alpha-lipoic acid in liver metabolism anddisease. Free. Radic. Biol. Med., 24, 1023–1039.

2.Estrada,D.E. et al. (1996) Stimulation of glucose uptake by the naturalcoenzyme alpha-lipoic acid/thioctic acid: participation of elements of theinsulin signaling pathway. Diabetes, 45, 1798–1804.

3.Konrad,D. et al. (2001) The antihyperglycemic drug alpha-lipoic acid stim-ulates glucose uptake via both GLUT4 translocation and GLUT4 activa-tion: potential role of p38 mitogen-activated protein kinase in GLUT4activation. Diabetes, 50, 1464–1471.

4. Jacob,S. et al. (1995) Enhancement of glucose disposal in patients withtype 2 diabetes by alpha-lipoic acid. Arzneimittelforschung, 45, 872–874.

5.Da Ros,R. et al. (2005) Molecular targets of diabetic vascular complica-tions and potential new drugs. Curr. Drug Targets, 6, 503–509.

6.Ziegler,D. (2004) Thioctic acid for patients with symptomatic diabeticpolyneuropathy: a critical review. Treat Endocrinol., 3, 173–189.

7.Haak,E. et al. (2000) Effects of alpha-lipoic acid on microcirculationin patients with peripheral diabetic neuropathy. Exp. Clin. Endocrinol.Diabetes, 108, 168–174.

8.Teichert,J. et al. (2003) Plasma kinetics, metabolism, and urinary excretionof alpha-lipoic acid following oral administration in healthy volunteers.J. Clin. Pharmacol., 43, 1257–1267.

9.Zhang,W.J. et al. (2001) Alpha-lipoic acid inhibits TNF-alpha-induced NF-kappaB activation and adhesion molecule expression in human aorticendothelial cells. FASEB J., 15, 2423–2432.

10.Bergers,G. et al. (2003) Tumorigenesis and the angiogenic switch. Nat. Rev.Cancer, 3, 401–410.

11.Tosetti,F. et al. (2002) Angioprevention’: angiogenesis is a common andkey target for cancer chemopreventive agents. FASEB J., 16, 2–14.

12.Albini,A. et al. (2001) Inhibition of angiogenesis-driven Kaposi’s sarcomatumor growth in nude mice by oral N-acetylcysteine. Cancer Res., 61,8171–8178.

13.Cai,T. et al. (1999) N-acetylcysteine inhibits endothelial cell invasion andangiogenesis. Lab Invest, 79, 1151–1159.

14.Fassina,G. et al. (2004) Mechanisms of inhibition of tumor angiogenesisand vascular tumor growth by epigallocatechin-3-gallate. Clin. CancerRes., 10, 4865–4873.

15.Albini,A. et al. (2006) Mechanisms of the antiangiogenic activity by thehop flavonoid xanthohumol: NF-kappaB and Akt as targets. FASEB J., 20,527–529.

16.Albini,A. et al. (2004) The ‘‘chemoinvasion assay’’: a tool to study tumorand endothelial cell invasion of basement membranes. Int. J. Dev. Biol., 48,563–571.

17.Benelli,R. et al. (1999) In vitro models of angiogenesis: the use of Matrigel.Int. J. Biol. Markers, 14, 243–246.

18.Albini,A. et al. (1997) The beta-core fragment of human chorionic gonad-otrophin inhibits growth of Kaposi’s sarcoma-derived cells and a newimmortalized Kaposi’s sarcoma cell line. AIDS, 11, 713–721.

19.Passaniti,A. et al. (1992) A simple, quantitative method for assessingangiogenesis and antiangiogenic agents using reconstituted basementmembrane, heparin, and fibroblast growth factor. Lab Invest, 67, 519–528.

20.Albini,A. et al. (1994) Angiogenic potential in vivo by Kaposi’s sarcomacell-free supernatants and HIV-1 tat product: inhibition of KS-like lesionsby tissue inhibitor of metalloproteinase-2. AIDS, 8, 1237–1244.

21.Pfeffer,U. et al. (2005) Molecular mechanisms of action of angiopreventiveanti-oxidants on endothelial cells: microarray gene expression analyses.Mutat. Res., 591, 198–211.

22.Singh,R.P. et al. (2005) Silibinin strongly inhibits growth and survival ofhuman endothelial cells via cell cycle arrest and downregulation of survi-vin, Akt and NF-kappaB: implications for angioprevention and antiangio-genic therapy. Oncogene, 24, 1188–1202.

23.Trigona,W.L. et al. (2006) Thioredoxin reductase regulates the induction ofhaem oxygenase-1 expression in aortic endothelial cells. Biochem. J., 394,207–216.

24.Kaga,S. et al. (2005) Resveratrol enhances neovascularization in the in-farcted rat myocardium through the induction of thioredoxin-1, heme oxy-genase-1 and vascular endothelial growth factor. J. Mol. Cell. Cardiol., 39,813–822.

25.Dulak,J. et al. (2004) Complex role of heme oxygenase-1 in angiogenesis.Antioxid. Redox Signal., 6, 858–866.

26.Morkunaite,S. et al. (2000) Mechanism of dihydrolipoate stimulation of themitochondrial permeability transition: effect of different respiratory sub-strates. IUBMB Life, 49, 211–216.

27.Wenzel,U. et al. (2005) Alpha-lipoic acid induces apoptosis in human coloncancer cells by increasing mitochondrial respiration with a concomitantO2-.-generation. Apoptosis, 10, 359–368.

28.Yonenaga,Y. et al. (2005) Absence of smooth muscle actin-positive peri-cyte coverage of tumor vessels correlates with hematogenous metastasisand prognosis of colorectal cancer patients. Oncology, 69, 159–166.

29.Hellstrom,M. et al. (2001) Lack of pericytes leads to endothelial hyperpla-sia and abnormal vascular morphogenesis. J. Cell Biol., 153, 543–553.

30. Jain,R.K. (2005) Normalization of tumor vasculature: an emerging conceptin antiangiogenic therapy. Science, 307, 58–62.

31.Morini,M. et al. (1999) The role of the thiol N-acetylcysteine in theprevention of tumor invasion and angiogenesis. Int. J. Biol. Markers, 14,268–271.

32.Garbisa,S. et al. (1999) Tumor invasion: molecular shears blunted by greentea. Nat. Med., 5, 1216.

33.Gupta,S. et al. (2003) Molecular pathway for (-)-epigallocatechin-3-gallate-induced cell cycle arrest and apoptosis of human prostate carcinomacells. Arch. Biochem. Biophys., 410, 177–185.

34.Packer,L. et al. (1996) Alpha-lipoic acid: the metabolic antioxidant. FreeRadic. Biol. Med., 20, 625–626.

35.Morikawa,T. et al. (2001) Do mammalian cells synthesize lipoic acid?Identification of a mouse cDNA encoding a lipoic acid synthase locatedin mitochondria. FEBS Lett., 498, 16–21.

36.Kowluru,R.A. et al. (2004) Effect of long-term administration of alpha-lipoic acid on retinal capillary cell death and the development of retino-pathy in diabetic rats. Diabetes, 53, 3233–3238.

37.Marsh,S.A. et al. (2005) Bcl-2 in endothelial cells is increased by vitamin Eand alpha-lipoic acid supplementation but not exercise training. J. Mol.Cell. Cardiol., 38, 445–451.

38.Byun,C.H. et al. (2005) Alpha-lipoic acid inhibits TNF-alpha-inducedapoptosis in human bone marrow stromal cells. J. Bone Miner. Res., 20,1125–1135.

39.van de Mark,K. et al. (2003) Alpha-lipoic acid induces p27Kip-dependentcell cycle arrest in non-transformed cell lines and apoptosis in tumor celllines. J. Cell. Physiol., 194, 325–340.

40.Nowis,D. et al. (2006) Heme oxygenase-1 protects tumor cellsagainst photodynamic therapy-mediated cytotoxicity. Oncogene, 25,3365–3374.

41. Jazwa,A. et al. (2006) Effect of heme and heme oxygenase-1 on vascularendothelial growth factor synthesis and angiogenic potency of humankeratinocytes. Free Radic. Biol. Med., 40, 1250–1263.

42.Yoo,M.H. et al. (2006) Thioredoxin reductase 1 deficiency reverses tumorphenotype and tumorigenicity of lung carcinoma cells. J. Biol. Chem., 281,13005–13008.

43.Loboda,A. et al. (2005) Heme oxygenase-1-dependent and -independentregulation of angiogenic genes expression: effect of cobalt protoporphyrinand cobalt chloride on VEGF and IL-8 synthesis in human microvascularendothelial cells. Cell. Mol. Biol. (Noisy-le-grand), 51, 347–355.

44.Bussolati,B. et al. (2004) Bifunctional role for VEGF-induced heme oxy-genase-1 in vivo: induction of angiogenesis and inhibition of leukocyticinfiltration. Blood, 103, 761–766.

45.Deramaudt,B.M. et al. (1998) Gene transfer of human heme oxygenaseinto coronary endothelial cells potentially promotes angiogenesis. J. Cell.Biochem., 68, 121–127.


1019


46.McAllister,S.C. et al. (2004) Kaposi sarcoma-associated herpesvirus(KSHV) induces heme oxygenase-1 expression and activity in KSHV-infected endothelial cells. Blood, 103, 3465–3473.

47.Marinissen,M.J. et al. (2006) Inhibition of heme oxygenase-1 interfereswith the transforming activity of the Kaposi sarcoma herpesvirus-encodedG protein-coupled receptor. J. Biol. Chem., 281, 11332–11346.

48.Phillips,D.J. et al. (2001) Evidence for activin A and follistatin involve-ment in the systemic inflammatory response. Mol. Cell. Endocrinol., 180,155–162.

49.Breit,S. et al. (2000) The N-myc oncogene in human neuroblastoma cells:down-regulation of an angiogenesis inhibitor identified as activin A.Cancer Res., 60, 4596–4601.

50.Griendling,K.K. et al. (2000) Modulation of protein kinase activity and geneexpression by reactive oxygen species and their role in vascular physiologyand pathophysiology. Arterioscler. Thromb. Vasc. Biol., 20, 2175–2183.

51.Alleva,R. et al. (2005) Alpha-lipoic acid supplementation inhibits oxidativedamage, accelerating chronic wound healing in patients undergoing hyper-baric oxygen therapy. Biochem. Biophys. Res. Commun., 333, 404–410.

52.Savitha,S. et al. (2005) Oxidative stress on mitochondrial antioxidantdefense system in the aging process: role of DL-alpha-lipoic acid andL-carnitine. Clin. Chim. Acta, 355, 173–180.

53.Cakatay,U. et al. (2005) Prooxidant activities of alpha-lipoic acid onoxidative protein damage in the aging rat heart muscle. Arch. Gerontol.Geriatr., 40, 231–240.

54.Bhatti,F. et al. (2005) Mechanisms of antioxidant and pro-oxidant effects ofalpha-lipoic acid in the diabetic and nondiabetic kidney. Kidney Int., 67,1371–1380.

Received June 9, 2006; revised October 17, 2006;accepted November 17, 2006

P.Larghero et al.

1020


Biological assays and genomic analysis reveal lipoic acid ...

Documents

Transcript of Biological assays and genomic analysis reveal lipoic acid ...