inhibition of angiogenesis by selective estrogen receptor modulators 2015

Original Articles

Inhibition of angiogenesis by selective estrogen receptor modulatorsthrough blockade of cholesterol tracking rather than estrogenreceptor antagonismJoong Sup Shim a,b,*, Ruo-Jing Li b, Junfang Lv a, Sarah Head b, Eun Ju Yang a, Jun O. Liu b,**a Faculty of Health Sciences, University of Macau, Av. Universidade, Taipa, Macau SAR, Chinab Department of Pharmacology and Molecular Sciences, Johns Hopkins School of Medicine, 725 N Wolfe St, Baltimore, MD 21205, USA

A R T I C L E I N F O

Article history:Received 29 December 2014Received in revised form 16 March 2015Accepted 17 March 2015

Keywords:AngiogenesisSelective estrogen receptor modulatorTamoxifenCholesterol tracking

A B S T R A C T

Selective estrogen receptor modulators (SERM) including tamoxifen are known to inhibit angiogenesis.However, the underlying mechanism, which is independent of their action on the estrogen receptor (ER),has remained largely unknown. In the present study, we found that tamoxifen and other SERM inhib-ited cholesterol tracking in endothelial cells, causing a hyper-accumulation of cholesterol in lateendosomes/lysosomes. Inhibition of cholesterol tracking by tamoxifen was accompanied by abnor-mal subcellular distribution of vascular endothelial growth factor receptor-2 (VEGFR2) and inhibition ofthe terminal glycosylation of the receptor. Tamoxifen also caused perinuclear positioning of lysosomes,which in turn trapped the mammalian target of rapamycin (mTOR) in the perinuclear region of endo-thelial cells. Abnormal distribution of VEGFR2 and mTOR and inhibition of VEGFR2 and mTOR activitiesby tamoxifen were signicantly reversed by addition of cholesterolcyclodextrin complex to the culturemedia of endothelial cells. Moreover, high concentrations of tamoxifen inhibited endothelial and breastcancer cell proliferation in a cholesterol-dependent, but ER-independent, manner. Together, these resultsunraveled a previously unrecognized mechanism of angiogenesis inhibition by tamoxifen and other SERM,implicating cholesterol tracking as an attractive therapeutic target for cancer treatment.

2015 Published by Elsevier Ireland Ltd.

Introduction

Tamoxifen and selective estrogen receptor modulators (SERM)have been used to treat hormone responsive, estrogen receptor (ER)-positive breast cancers since the 1980s. It has generally been acceptedthat the anticancer activity of tamoxifen is mainly attributable toits competitive antagonism to ER, thereby inhibiting the prolifer-ation of ER-positive breast cancer cells [1]. However, whether thisis the only mechanism of action underlying the anticancer activi-ty of SERM has been questioned since tamoxifen and other SERMalso showed anticancer activity in ER-negative breast cancers [24].Since the 1990s, several groups have found that tamoxifen and SERMstrongly inhibited angiogenesis by mechanisms independent of ER[57]. Based on these ndings, tamoxifen and other SERM are nowbeing actively investigated as anti-angiogenic agents in clinical trialsfor cancer treatment [810]. However, the underlying molecularmechanism by which tamoxifen inhibits angiogenesis has re-mained largely unknown.

Cholesterol is an essential component of cellular membranes andplays a key role in membrane permeability and uidity. In addi-tion to a structural role, it also functions in intracellular transportand cell signaling [11,12]. Serum cholesterol is delivered through-out the body in the form of low-density lipoprotein (LDL) andtransported into cells through receptor-mediated endocytosis [13].Endocytosed LDL is transported to the late endosomes and lyso-somes (endolysosomes) where cholesteryl esters are hydrolyzed andfree cholesterol is released from the endosomal system for deliv-ery to other compartments, including the plasma membrane andendoplasmic reticulum [14]. One of the most important machin-eries of cholesterol tracking in the endolysosomes is the NiemannPick type C (NPC) proteins (NPC1 and NPC2), which help acid lipase-mediated hydrolysis of cholesteryl esters and deliver free cholesterolout of the endolysosomes [15]. Inhibition of NPC1 or 2 causes ac-cumulation of cholesterol and glycolipids in the endolysosomes, aphenotype called NPC after the genetic disease of the same name[16].

We have previously reported that a newly identied anti-angiogenic drug itraconazole inhibited cholesterol tracking andinduced NPC-like phenotype in endothelial cells [17]. Inhibition ofcholesterol tracking by itraconazole is accompanied by inhibi-tion of mTOR signaling and VEGFR2 glycosylation, both of whichare essential signaling components for endothelial cell proliferation

* Corresponding author. Tel.: +853 8822 4990.E-mail address: [email protected] (J.S. Shim).

** Corresponding author. Tel.: +1 410 955 4619.E-mail address: [email protected] (J.O. Liu).

http://dx.doi.org/10.1016/j.canlet.2015.03.0220304-3835/ 2015 Published by Elsevier Ireland Ltd.

Cancer Letters (2015)

ARTICLE IN PRESS

Please cite this article in press as: Joong Sup Shim, et al., Inhibition of angiogenesis by selective estrogen receptor modulators through blockade of cholesterol trafficking rather thanestrogen receptor antagonism, Cancer Letters (2015), doi: 10.1016/j.canlet.2015.03.022

Contents lists available at ScienceDirect

Cancer Letters

journal homepage: www.elsevier.com/ locate /canlet

Q2

Q1

123

4

567

89

1011121314151617181920212223242526272829303132333435363738

39

404142434445464748495051525354555657585960616263

646566676869707172737475767778798081828384858687

[17,18]. Recently, Fang et al. showed that upon over-expression,apoA-I binding protein (AIBP), which is responsible for cholesteroleux from endothelial cells, inhibited angiogenesis by depletingcholesterol from the plasma membrane, thereby inhibiting theVEGFR2 signaling pathway in endothelial cells and animal models[19]. Similar to AIBP over-expression, cells with NPC phenotypeinduced by small molecules showed accumulation of cholesterol inthe endolysosomes leading to cholesterol depletion in plasmamem-brane and inhibition of the VEGFR2 signaling pathway [17,18]. Theseresults strongly suggest that cholesterol tracking in endothelialcells is critical for proper angiogenesis.

In the present study, we found that tamoxifen and other SERMinhibited cholesterol tracking in endothelial cells. Blockade of cho-lesterol trackingbySERMled toanabnormal subcellulardistributionof mTOR and VEGFR2 and caused inhibition of their signaling path-ways in a cholesterol-dependent manner. These data suggest thattamoxifen and other SERM inhibit angiogenesis by interfering withcholesterol tracking in endothelial cells and that cholesterol traf-cking is a novel target for anti-angiogenesis therapy.

Materials and methods

Cells and reagents

Pooled human umbilical vein endothelial cells (HUVEC) were purchased fromLonza (Allendale, NJ) and were grown in endothelial cell growth medium-2 (EGM-2)using the EGM-2 bullet kit (Lonza). MCF-7 (ER-positive) breast cancer cells were grownin Roswell Park Memorial Institute (RPMI)-1640medium containing 10% fetal bovineserum (FBS, Life Technologies, Grand Island, NY) and 1% antibiotics (penicillin andstreptomycin) solution (Life Technologies). MDA-MB-231 (triple negative) breast cancercells were grown in high-glucose Dulbeccos Modied Eagles Medium (DMEM) with10% FBS (Life Technologies) and 1% antibiotics. All the cells were maintained in ahumidied incubator at 37 C adjusted to 5% CO2. Methyl--cyclodextrin, choles-terol and lipin were purchased from Sigma-Aldrich (St. Louis, MO). Recombinanthuman VEGF-165 was purchased from R&D Systems (Minneapolis, MN).

Filipin staining

Filipin staining was performed as describedwith slight modications [17]. HUVECwere cultured in a Nunc Lab-Tek II 8-Chamber Slide (Thermo Scientic, Rockford,IL) at 1 104 cells/well. Cells were treated with SERM with or without cholesteroland cyclodextrin complex for 24 h. Cells were then xed with 4% paraformalde-hyde for 20min at room temperature and stained with lipin at a nal concentrationof 50 g/ml in the dark for 1 h at room temperature. Cells were washed with PBS,mounted with Immu-mount (Thermo Scientic), and observed under a Zeiss 510Meta multiphoton confocal microscope (Carl Zeiss, Thornwood, NY).

Immunouorescence imaging

For co-staining of proteins and cholesterol, HUVEC (1 104 cells/well) grown ina Nunc Lab-Tek II 8-Chamber Slide were treated with compounds for 24 h, xed with4% paraformaldehyde for 20 min at room temperature and stained with lipin(50 g/ml) for 1 h at room temperature. Cells were then permeabilized with 0.2%saponin supplemented with 50 g/ml lipin and 5% bovine serum albumin (BSA)in PBS for 30min. Cells were incubated with primary antibodies in PBS together with50 g/ml lipin, 0.05% saponin and 5% BSA overnight at 4 C. Cells were then incu-bated with secondary antibodies in PBS with 50 g/ml lipin, 0.05% saponin and 5%BSA at room temperature for 1 h. Cells were washed with PBS, mounted with Immu-mount, and observed under a Zeiss 510 Meta multiphoton confocal microscope. Forgeneral immunouorescence, cells were xed with 4% paraformaldehyde for 10min,permeabilized with 0.5% Triton X-100 for 10min and washed with PBS prior to block-ing in 1% BSA in PBS containing 0.1% Tween 20 (PBST) for 1 h. Cells were thenincubated with primary antibodies including anti-VEGFR2 (Cell Signaling Technol-ogy, Danvers, MA), anti-LAMP1 (Santa Cruz Biotechnology, Santa Cruz, CA), anti-mTOR (Cell Signaling Technology) and anti-GM130 (BD Biosciences, San Jose, CA)in the blocking solution overnight at 4 C, and then incubated with secondary an-tibodies conjugated with Alexa-Fluo488 or Alexa-Fluo594 for 1 h. The cellular nucleiwere stained with 4,6-diamidino-2-phenylindole (DAPI) and actin cytoskeleton wasstained with rhodamine-phalloidin (Life Technologies). The immunouorescenceimages were obtained using the Zeiss 510 Meta multiphoton confocal microscope.

[3H]-thymidine DNA incorporation assay

HUVEC, MCF-7 orMDA-MB-231were seeded at 3 103 cells/well in 96-well platesand allowed to adhere at 37 C for 24 h. Cells were then treated with compoundsfor 24 h prior to being pulsed with 0.5 Ci [3H]-thymidine (PerkinElmer, Waltham,

MA) for 16 h and then trypsinized. The cells were harvested onto FilterMat A glassber lters (Wallac, Turku, Finland) using a Harvester 96 cell harvester (Tomtec,Hamden, CT), and the radioactivity of [3H]-thymidine incorporated into DNA wascounted using a MicroBeta liquid scintillation plate reader (PerkinElmer). The IC50values and 95% condence intervals were calculated using the GraphPad Prism 5.0software (GraphPad Software, San Diego, CA).

AlamarBlue cell viability assay

HUVEC, MCF-7 orMDA-MB-231were seeded at 3 103 cells/well in 96-well platesand allowed to adhere at 37 C for 24 h. The cells were treated with compounds inthe presence or absence of cholesterol or cyclodextrin, or both for 24 h. AlamarBluereagent (Life Technologies) was added to the media at a nal concentration of 10%and the incubation was continued for an additional 2 h. The uorescence signal wasread with an excitation wavelength of 570 nm and an emission wavelength of 590 nmusing a SpectraMax M5 uorescence microplate reader (Molecular Devices, Sunny-vale, CA).

Western blot analysis

HUVEC (2 105 cells/well) were seeded in 6-well plates and allowed to adhereovernight. Following drug treatment for 24 h, cells were lysed by adding 2 Laemmlibuffer containing 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.004% bromophe-nol blue, 0.125M TrisHCl, pH 6.8 and the lysates were boiled for 10min and vortexed.After SDS-PAGE, the proteins were transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA). The blots were blocked with 5% non-fat dried milk at roomtemperature for 1 h, incubated with the primary antibodies including anti-VEGFR2(Cell Signaling Technology), anti-phospho-VEGFR2 (Tyr1175, Cell Signaling Tech-nology), anti-FGFR1 (Cell Signaling Technology), anti-mTOR (Cell Signaling Technology),anti-phospho-mTOR (Ser2448, Cell Signaling Technology), anti-S6K (Cell SignalingTechnology), anti-phospho-S6K (Thr389, Cell Signaling Technology), anti-PDGFR(Santa Cruz Biotechnologies), anti-actin (Santa Cruz Biotechnologies) or anti--tubulin (Santa Cruz Biotechnologies) antibodies overnight at 4 C and then incubatedwith HRP-conjugated secondary antibodies at room temperature for 1 h. The immune-complexes were detected using enhanced chemiluminescence (ECL) detection reagent(GE Healthcare, Pittsburgh, PA).

Endothelial cell tube formation assay

A 96-well plate was coated with 50 l Matrigel (BD Biosciences) and was incu-bated at 37 C for 1 h to allow for polymerization. HUVECweremixedwith appropriatecompounds and seeded (2 104 cells/well) on the Matrigel-coated wells, followedby incubation in a CO2 incubator for 16 h. Cells were washed carefully with PBS onceand Calcein-AM (2 M in PBS, Life Technologies) solution was added to the cells.After incubation at 37 C for 30 minutes, the cells were washed gently with PBS andthe uorescence-labeled tubular structures were observed under a Nikon EclipseTS100 uorescence microscope with an excitation wavelength of 485 nm and anemission wavelength of 520 nm at magnication 100. The total tube lengths, sizesand number of junctions from the uorescence images were quantied using theAngioQuant v1.33 software (The MathWorks, Natick, MA) and plotted using theGraphPad Prism 5.0 software.

Statistical analysis

Statistical signicance of the data between control and test groups was deter-mined by two-sided Students t-test using the GraphPad Prism 5.0 software. TheP values less than 0.05 were considered signicant.

Results

Tamoxifen and SERM induce NPC-like phenotype in HUVEC

We tested four FDA-approved SERM including tamoxifen,toremifene, clomifene and raloxifene (Supplementary Fig. S1) foreffects on cholesterol tracking in HUVEC. Intracellular cholester-ol was visualized by staining xed cells with lipin [20]. All fourSERM induced accumulation of cholesterol in the perinuclear regionof HUVEC, a phenotype similar to NPC and that induced byitraconazole (Fig. 1A and Supplementary Fig. S2A). The cell mor-phology was also changed from large at morphology into a partiallyshrunken form. These effects, however, were reversed upon addi-tion of exogenous cholesterol and its carrier methyl--cyclodextrin(CD) (Fig. 1B and Supplementary Fig. S2B), suggesting that they aremediated largely through inhibition of cholesterol tracking.

ARTICLE IN PRESS


2 J.S. Shim et al./Cancer Letters (2015)

123456789

10111213141516171819

20

21

2223242526272829303132

33

3435363738394041

42

434445464748495051525354555657585960616263

64

656667

6869707172737475

76777879808182838485

8687888990919293949596979899100101102103

104105106107108109110111112113114115116117

118119120121122123124125126127128129130131132133134135136137

Tamoxifen traps VEGFR2 in Golgi and induces perinuclear positioningof lysosomes in HUVEC

We have previously reported that a cholesterol tracking inhib-itor, itraconazole, inhibited theVEGFR2andmTORsignalingpathways,both ofwhich have been shown to be critical for angiogenesis [17,18].To determine the effect of tamoxifen on VEGFR2 and mTOR signal-ing,werst analyzed the subcellular localizationof those twoproteins.Immunouorescence labeling showed that VEGFR2 was localizedlargely in the plasmamembrane and some in intracellular regions incontrol HUVEC (Fig. 2A). Treatment of cells with tamoxifen causedan accumulation of a signicant amount of VEGFR2 in the peri-nuclear region,whichcolocalizedwith theGolgimarkerGM130(Fig. 2AandB). Todetermine if theabnormal subcellulardistributionofVEGFR2caused by tamoxifen is due to the inhibition of cholesterol track-ing,weaddedcholesterol/CDcomplexto thecell culturemedia.VEGFR2mislocalizationbytamoxifenwasclearly reversedbythecholesterol/CDcomplex (Fig. 2B). On the other hand, mTOR was mainly located inthe peripheral cytoplasm in controlHUVEC andwaswell co-localizedwith the endolysosomal marker, LAMP1 (Fig. 2C). Tamoxifen treat-ment caused cholesterol accumulation in the perinuclear regionoverlapping with the LAMP1-positive endolysosomes. Thoughtamoxifen did not alter mTOR association with LAMP1-positiveendolysosomes, the localization of both mTOR and lysosomes wasconned to the perinuclear region in the tamoxifen-treated HUVEC,whereas both proteins were more evenly distributed in the cyto-plasm in control HUVEC. Staining of the actin cytoskeleton clearlyshowed thatmTOR localizationwas conned to theperinuclear regionby tamoxifen treatment (Fig. 2D). This effect was reversed bycholesterol/CDcomplex(Fig. 2D).Theseresults indicatedthat tamoxifencaused subcellular redistribution of VEGFR2 and mTOR in endothe-lial cells in a cholesterol-dependent manner.

Tamoxifen inhibits VEGFR2 glycosylation and mTOR activity inHUVEC

Proper subcellular localization of proteins is critical for theirproper function. Like other cell surface proteins, VEGFR2 is highlyglycosylated and is expressed on the cell surface upon completion

of glycosylation. Protein glycosylation occurs co-translationally inthe endoplasmic reticulum and subsequently in the Golgi by a seriesof glycosidases and glycosyltransferases located in each organelle[21]. Abnormal subcellular localization of glycoproteins will causeimproper glycosylation and, therefore, affect their functions. Astamoxifen caused abnormal subcellular distribution of VEGFR2, wedetermined the effect of tamoxifen on the glycosylation pattern andthe activity of VEGFR2. Control HUVEC showed two glycosylatedforms of VEGFR2 (200 and 230 kD) (Fig. 3A). The 230 kD (matureterminal glycosylated form) protein band was dominant over the200 kD (intermediate glycosylated form) band. Tunicamycin is aglycosyltransferase inhibitor that acts in an initial step ofglycosylation in the endoplasmic reticulum. It completely inhib-ited VEGFR2 glycosylation, reducing the proteins mass to 180 kDas expected. Deoxymannojirimycin (dMM), an inhibitor of-mannosidases, and itraconazole are known to inhibit terminalglycosylation, thus shifting the VEGFR2 from 230 to 200 kD form.Similar to dMM and itraconazole, tamoxifen and other SERM causeda shift in the apparent molecular mass of VEGFR2 from 230 to200 kD, suggesting that they inhibited terminal glycosylation whichmainly occurs in the Golgi. Inhibition of terminal glycosylation ofVEGFR2 by SERM occurred in a concentration-dependent manner(Fig. 3BD).

Subcellular localization of mTOR is also important for its activ-ity. Recently, it was reported that nutrient starvation inducedabnormal lysosomal positioning (increase in perinuclear position-ing) [22]. This perinuclear lysosomal positioning was accompaniedby mTOR Complex-1 (mTORC1) redistribution and inhibited the ac-tivity of mTOR. Tamoxifen and other SERM also caused perinuclearlysosomal positioning (Fig. 2C and D) and inhibited the phosphory-lation of S6 kinase (S6K), a substrate of mTORC1, in HUVEC(Fig. 3BD). To further validate the inhibitory effect of tamoxifen onmTORC1 activity, the phosphorylation status of mTOR at Ser2448was assessed. Like the direct mTOR inhibitor rapamycin, tamoxifendose-dependently inhibited mTOR phosphorylation at Ser2448 inHUVEC (Fig. 4A). The phosphorylation of S6K was also inhibited bytamoxifen in parallel with the inhibition of mTOR phosphorylation.

We next examined glycosylation patterns of other receptor ty-rosine kinases related to angiogenesis, including broblast growth

Fig. 1. Effect of SERM on cholesterol tracking in HUVEC. (A) HUVEC were treated with tamoxifen (TMX), toremifene (TRM), clomifene (CLM) and raloxifene (RLX) for 24 h,and intracellular cholesterol was visualized by lipin. (B) HUVEC were treated with 1 M tamoxifen (TMX, upper panel) or 1 M toremifene (TRM, lower panel) in the pres-ence or absence of cholesterol (5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD) for 24 h, and intracellular cholesterol was visualized by lipin. Representative confocal imagesfrom four independent experiments are shown.

ARTICLE IN PRESS


3J.S. Shim et al./Cancer Letters (2015)

1234

5

6789

1011121314151617181920212223242526272829303132333435363738394041424344

454647484950515253545556575859606162636465666768697071727374757677787980818283

factor receptor 1 (FGFR1) and platelet-derived growth factor re-ceptor (PDGFR), to see if the tamoxifen effect was specic toVEGFR2. Similar to the effect observed on VEGFR2, tamoxifen causeda shift in the apparent molecular mass of PDGFR and FGFR1(Fig. 4B). Sorafenib, a kinase inhibitor, had no effect on the recep-tor molecular masses. However dMM caused a shift in the receptorsmolecular masses in a manner similar to that of tamoxifen, sug-gesting that tamoxifen inhibited terminal glycosylation of thereceptor tyrosine kinases. These results suggested that tamoxifeninhibits a common pathway in terminal glycosylation of receptortyrosine kinases in endothelial cells.

Cholesterol reverses the inhibitory effects of tamoxifen on VEGFR2and mTOR signaling in HUVEC

We next determined the effect of cholesterol on the inhibitionof VEGFR2 glycosylation by tamoxifen in HUVEC. Inhibition of ter-minal glycosylation by either tamoxifen or toremifene wascompletely reversed by the addition of cholesterol/CD complex

(Fig. 5A and B). To see if the inhibition of terminal glycosylation bytamoxifen affected VEGFR2 activity/signaling, tyrosine phosphory-lation status of VEGFR2 was assessed. Inhibition of terminalglycosylation of VEGFR2 by tamoxifen led to the inhibition of VEGFR2phosphorylation which was completely reversed by cholesterol/CD complex (Fig. 5C). A VEGFR2 tyrosine kinase inhibitor sunitinibalso inhibited VEGFR2 phosphorylation. But this effect was not re-versed by cholesterol/CD complex (Fig. 5C). These data suggestedthat the effect of tamoxifen on VEGFR2 was mediated through in-hibition of cholesterol tracking in endothelial cells. We furtherinvestigated the effect of cholesterol on inhibition of mTOR signal-ing by tamoxifen in HUVEC. Inhibition of S6K phosphorylation byeither tamoxifen or toremifene was completely reversed bycholesterol/CD complex (Fig. 5D). CD alone could partially reversetamoxifen activity. This was presumably due to the reversal effectof CD on NPC phenotype by releasing free cholesterol fromendolysosomes [23]. Inhibition of mTOR signaling by tamoxifen andits reversal by cholesterol were further conrmed by examining 4E-BP1 phosphorylation, which was also decreased (SupplementaryFig. S3).

Fig. 2. Effects of SERM and cholesterol on the subcellular localization of VEGFR2 and mTOR in HUVEC. (A) HUVEC were treated with or without 1 M tamoxifen (TMX) for24 h and subcellular localization of VEGFR2 was assessed under a confocal microscope. Arrows indicate VEGFR2. (B) HUVEC were treated with or without 1 M tamoxifen(TMX) in the presence or absence of cholesterol (5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD) for 24 h and subcellular localization of VEGFR2 and Golgi (GM130) wasassessed. (C) HUVEC were treated with or without 1 M tamoxifen (TMX) for 24 h and subcellular localization of cholesterol (Filipin), lysosomes (LAMP1), and mTOR wasassessed. (D) HUVEC were treated with 1 M tamoxifen (TMX) in the presence or absence of cholesterol (5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD) for 24 h and sub-cellular localization of mTOR and actin was assessed. Arrows indicate that tamoxifen induced a change in mTOR localization from peripheral cytoplasm to the perinuclearregion and this was reversed by cholesterolcyclodextrin complex. Representative confocal images from four independent experiments are shown.

ARTICLE IN PRESS



1234567

8

910111213141516171819

20

21222324252627

2829303132333435363738394041424344454647

Cholesterol reverses the inhibitory effects of tamoxifen on HUVECproliferation

To assess the relationship between VEGFR2 phosphorylation,mTOR activity and cell proliferation, we examined the effect of SERMonHUVEC proliferation. Half maximal inhibitory concentrations (IC50)of tamoxifen, toremifene and raloxifene for HUVEC proliferationweredetermined to be 0.98, 1.2, and 1.42 M, respectively (Fig. 6A andSupplementary Table S1). We then determined if cholesterol couldreverse the inhibition of HUVEC proliferation by SERM. Tamoxifenand toremifene strongly inhibited the growth of HUVEC at 2 and4 M, respectively. The inhibitions were partially reversed by CD andwere fully reversed by cholesterol/CD complex (Fig. 6B). Cholester-ol alone has no reversal effect on the cell growth inhibition bytamoxifen or toremifene. These data corroborated the effects of SERMon VEGFR2 and mTOR (Fig. 5AD).

Inhibition of cholesterol tracking by tamoxifen is independent of ER

SERM are potent antagonists of ER, with Kd values ranging frompicomolar to single-digit nanomolar concentrations, hence showinganti-proliferative effects on ER-positive breast cancer cells [24]. Wethus tested whether SERM have different sensitivity on cell prolif-eration and cholesterol tracking in cells with different ERexpression statuses. In an ER-positive cell line (MCF-7), SERM showeda biphasic cell growth inhibition; marginal inhibition at lower con-centrations (from nanomolar to single-digit micromolar, dottedarrows) and strong/complete inhibition at higher concentrations(from single- to double-digit micromolar, solid arrows) (Fig. 7A). Inan ER-negative cell line (MDA-MB-231), SERM showed a typicaldoseresponse curve with growth inhibition at higher concentra-tions (from single- to double-digit micromolar) (Fig. 7B). These datasuggested that SERM have at least two independent targets for

Fig. 3. Effect of SERM on VEGFR2 glycosylation and mTORC1 pathway in HUVEC. (A) HUVEC were treated with SERM including tamoxifen (TMX, 5 M), toremifene (TRM,5 M) and clomifene (CLM, 5 M) for 24 h and VEGFR2 glycosylation was assessed byWestern blotting. Glycosylation inhibitors including deoxymannojirimycin (dMM, 500 M)and tunicamycin (TUM, 2 g/ml), and a cholesterol tracking inhibitor itraconazole (ITRA, 1 M) were used as positive controls. Three different glycosylated forms of VEGFR2(a: 230 kD hyper-glycosylated form, b: 200 kD intermediate glycosylated form and c: 180 kD unglycosylated form) are shown. (BD) Effect of various concentrations of SERMon VEGFR2 glycosylation and mTORC1 pathway indicated by the level of phosphorylated S6K (pS6K) are shown. Representative Western blot images from three inde-pendent experiments are shown.

Fig. 4. Effect of tamoxifen on the phosphorylation of mTOR and the glycosylation of receptor tyrosine kinases in HUVEC. (A) HUVEC were treated with tamoxifen (TMX) orrapamycin (Rapa) at indicated concentrations for 24 h and the phosphorylation of mTOR at Ser2448 as well as phosphorylated S6K (pS6K) and total S6K was analyzed.(B) HUVEC were treated with tamoxifen (TMX), sorafenib (Soraf) or deoxymannojirimycin (dMM) at indicated concentrations for 24 h. The terminal glycosylation of thereceptor tyrosine kinases was assessed by Western blotting using specic antibodies against each receptor tyrosine kinase in the presence of the known glycosylation in-hibitor dMM. Representative Western blot images from three independent experiments are shown.

ARTICLE IN PRESS



123456

7

89

1011121314151617181920212223

24

2526272829

30313233343536373839404142434445

growth inhibition inMCF-7 cells. It could be postulated that themoresensitive target is ER while the less sensitive target is cholesteroltracking in ER-positive MCF-7 cells. We further tested the effectof a high concentration of tamoxifen (10 M) on cholesterol traf-cking in both MCF-7 and MDA-MB-231. Tamoxifen stronglyinhibited cholesterol tracking in both cells and the inhibition wasreversed by cholesterol/CD complex (Fig. 7C). A high concentra-tion of tamoxifen strongly inhibited the cell growth of both MCF-7and MDA-MB-231 and the inhibition was reversed by cholesterol/CD complex (Fig. 7D). These data demonstrated that inhibition ofcholesterol tracking by tamoxifen is independent of its action onER.

Cholesterol reverses the inhibitory effect of tamoxifen on HUVEC tubeformation in Matrigel

We determined the effect of SERM on endothelial cell tube for-mation, a well-established in vitro assessment of angiogenesis. Asexpected from previous reports [5,6], both tamoxifen and toremifeneinhibited the tube formation of HUVEC on Matrigel (Fig. 8A). Theinhibition of tube formation by SERM, however, was signicantlyreversed by an addition of cholesterol/CD complex, while cholesterol/CD itself did not affect the tube structures (Fig. 8A and B). Takentogether, these results suggested that anti-angiogenic activity oftamoxifen is mainly mediated by its effect on cholesterol track-ing in endothelial cells.

Discussion

SERM are mixed agonists/antagonists of ER, which act differ-ently depending on cell and tissue types. As they act as antagonistsof ER in breast tissue, SERM, especially tamoxifen, have long beenused to treat ER-positive breast cancer [25]. However, it has beenquestioned whether the anticancer effect of SERM is solely due tothe ER antagonism, since a number of reports have shown the

therapeutic effects of tamoxifen on ER-negative breast cancer [24].The anti-angiogenic activity of SERMwas rst reported by Gagliardiand Collins in 1993 [6]. SERM including clomiphene and tamoxifensignicantly inhibited angiogenesis in the chorioallantoic mem-brane in growing chick embryos. Addition of excessive amount of17-estradiol did not alter the anti-angiogenic activity of SERM, sug-gesting that angiogenesis inhibition by SERM was independent ofthe blockade of ER. Subsequent studies have demonstrated that SERMare effective anti-angiogenic agents using several animal models,including ER-negative rat models [5,7]. Based on these observa-tions, SERM are under review in several clinical studies as an anti-angiogenic monotherapy or an adjuvant therapy with chemotherapydrugs in a broad range of cancer types [810]. All the evidencestrongly suggests that inhibition of angiogenesis is one of the majormechanisms underlying the anticancer activity of SERM in addi-tion to ER. However, the mechanism underlying the anti-angiogenicactivity of SERM has remained elusive.

In the present study, we showed that SERM inhibited choles-terol tracking in endothelial cells, as evidenced by abnormalaccumulation of free cholesterol in the endolysosomes. This effectwas accompanied by aberrant subcellular localization of two majorplayers in angiogenesis, VEGFR2 and mTOR, in endothelial cells.VEGFR2 undergoes glycosylation upon translation and theglycosylation is required for receptor auto-phosphorylation and ac-tivation [26]. Tamoxifen treatment caused trapping of the VEGFR2in the Golgi, inhibited the terminal glycosylation and depleted thereceptor in the plasma membrane. Consequently, tamoxifen inhib-ited the VEGF-induced phosphorylation of VEGFR2 in endothelialcells. Tamoxifen also inhibited terminal glycosylation of other re-ceptor tyrosine kinases related to angiogenesis such as FGFR1 andPDGFR in HUVEC. These data suggest that tamoxifen inhibits acommon terminal glycosylation pathway in endothelial cells leadingto the inhibition of maturation of the receptor tyrosine kinases. Onthe other hand, mTORC1 is recruited to the lysosome surface andregulates lysosomal functions when cells are in normal nutrient-rich

Fig. 5. Reversal effect of cholesterol on the inhibition of VEGFR2 and mTOR activities by SERM. (A and B) HUVEC were treated with tamoxifen (TMX) or toremifene (TRM)with or without cholesterol (5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD) for 24 h, and VEGFR2 glycosylation was assessed by Western blotting. (C) HUVEC were grownin low serum media (0.1% FBS without additional growth factor supplements) and treated with drugs including tamoxifen (TMX) and sunitinib (Sunit, 100 nM) for 24 h inthe presence or absence of cholesterol (5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD). Cells were then stimulated with 50 ng/ml of VEGF-165 for 5 min and the levels oftotal and phosphorylated VEGFR2 were assessed by Western blotting. (D) HUVEC were treated with tamoxifen (TMX, 5 M) or toremifene (TRM, 5 M) with or withoutcholesterol (Chol, 5 g/ml) and cyclodextrin (CD, 0.1%) for 24 h, and mTOR activity was assessed by Western blotting of phosphorylated S6-kinase (pS6K). RepresentativeWestern blot images from three independent experiments are shown.

ARTICLE IN PRESS



1234567

8

910111213141516171819202122232425262728293031323334353637383940414243

4445464748495051525354555657585960616263646566676869707172737475767778

state [27]. Conversely, mTOR activity is also regulated by lyso-somes by altering subcellular lysosomal positioning [22].The lysosomal positioning is critical for mTOR activity [22,28,29].Under starvation, intracellular pH (pHi) was increased and this pHichange in turn caused perinuclear clustering of lysosomes. Theperinuclear clustering of lysosomes led to an inhibition of mTORactivity. Forced movement of lysosomes to the cell periphery byoverexpressing kinesin family of proteins could restore mTOR ac-tivity, suggesting a critical role of lysosomal positioning in mTORsignaling pathway [22]. In our study, we found that tamoxifen didnot alter the association of mTOR with the LAMP1-positiveendolysosomes. Instead, it switched lysosomal positioning from thecell periphery to the perinuclear region leading to the inhibition ofmTOR activity. Although the mechanism by which lysosomal po-sitioning inuences mTOR activity remains to be elucidated, it hasbeen proposed that lysosomes in the cell periphery would enablemTOR to access to its upstream signaling molecules such as acti-vated Akt at the cell surface membrane. Conning lysosomes to theperinuclear region would prevent mTOR from accessing its up-stream signaling molecules at the cell membrane, leading to theblockade of its activation [22]. Further studies are necessary to elu-cidate the causal relationship between SERM-induced lysosomalpositioning and mTOR activity.

We further showed that addition of extracellular cholesterol couldsignicantly reverse the abnormal localization of VEGFR2 and in-hibition of terminal glycosylation and receptor phosphorylationcaused by SERM. Cholesterol also could reverse SERM-induced peri-nuclear lysosomal positioning and inhibition of mTOR activity. Thesedata strongly suggest that cholesterol tracking lies upstream ofVEGFR2 tracking and glycosylation and lysosomal positioning bySERM. In addition, inhibition of endothelial cell proliferation andtube formation by SERM was markedly reversed by the addition ofextracellular cholesterol, implying that inhibition of cholesterol traf-cking led to the inhibition of twomajor signaling pathways, VEGFR2and mTOR, which is likely the main mechanismmediating the anti-angiogenic activity of SERM.

In this study, we have not identied themolecular target respon-sible for the inhibition of endothelial cell cholesterol tracking bySERM. ER could be a candidate, but it was ruled out by testing twodifferent breast cancer cell lines, MCF-7 (ER-positive) andMDA-MB-231 (ER-negative). SERM at high concentrations caused abnormalcholesterol accumulation in both cell lines regardless of the ER statusand inhibited cell proliferation in a cholesterol-dependent manner.Onepossiblemechanismof cholesterol tracking inhibitionby SERMcould be that SERM, especially tamoxifen, act as a lipophilic weakbase and increase the pH in acidic organelles such as lysosomes [30].

Fig. 6. Effects of SERM and cholesterol on HUVEC proliferation. (A) HUVEC were treated with various concentrations of tamoxifen, toremifene or raloxifene for 24 h andcell proliferation was assessed through the [3]H-thymidine uptake assay. (B) HUVEC were treated with tamoxifen (TMX, 2 M) or toremifene (TRM, 4 M) for 24 h and wereobserved under a phase contrast microscope. Cholesterol (Chol, 5 g/ml) and cyclodextrin (CD, 0.1%) were added together to assess reversibility on the anti-proliferativeeffect of SERM. NT denotes not treated with Chol/CD. Representative phase-contrast images from four independent experiments are shown. The cell viability was quanti-ed by AlamarBlue staining and was plotted using the GraphPad Prism 5.0 software (right panel). Data represent mean standard deviation (SD) from four independentexperiments. **P < 0.01 between two indicated groups.

ARTICLE IN PRESS



123456

7

89

101112131415161718192021222324252627282930

3132333435363738394041424344454647484950515253

Changes in organellar pH could inhibit enzyme activities in the or-ganelles including NPC proteins and glycosylation enzymes, whichcould potentially affect cholesterol and receptor tyrosine kinase traf-cking in thecells.Ongoingstudiesare focusedon lysosomalpHchangein endothelial cells and the resultswill bepresented in thenear future.

The vascular endothelium is the rst layer of cells that are incontact with the full circulating lipoproteins from blood plasma, andis responsible for the uptake of LDL-cholesterol from the plasma andeux of cellular cholesterol to HDL in the plasma. Several recentreports showed that cholesterol uptake and eux in endothelial cellsare important regulators of angiogenesis [19,31,32]. Usui et al.showed that LDL alone could activate VEGFR1 signaling in theabsence of VEGF in macrophages [31]. The activation of VEGFR1 sig-naling was mediated by recruitment of the LDL receptor to VEGFR1by LDL during its uptake. Conversely, Fang et al. demonstrated thatAIBP mediates cholesterol eux from endothelial cells, and its

overexpression suppresses angiogenesis in animal models throughinhibition of VEGFR2 signaling [19]. In our previous studies,itraconazole was found to inhibit cholesterol tracking in endo-thelial cells [17]. Itraconazole is undergoing multiple clinical studiesas an anti-angiogenic agent. Several positive clinical results havebeen reported recently from Phase II studies for cancer treatment,including metastatic and castration-resistant prostate cancer, non-small cell lung cancer and basal cell carcinoma [3335]. These studiestogether with our current results suggest that endothelial cell cho-lesterol tracking can serve as a novel therapeutic target forangiogenesis-related diseases including cancer.

Acknowledgements

This work was supported by NIH/NCI (CA122814), FAMRI andProstate Cancer Foundation (to J.O.L) and by the Science and

Fig. 7. Effects of SERM and cholesterol on ER-positive or ER-negative breast cancer cell proliferation. (A and B) MCF-7 (ER-positive) or MDA-MB-231 (ER-negative) cells weretreated with various concentrations of tamoxifen, toremifene or raloxifene for 24 h and cell proliferation was assessed through the [3]H-thymidine uptake assay. SERM showeda biphasic growth inhibition in MCF-7 cells. Dotted arrows indicate concentration ranges that show marginal cell growth inhibition, whereas solid arrows represent con-centration ranges with strong cell growth inhibition. (C) MCF-7 or MDA-MB-231 cells were treated with 10 M tamoxifen (TMX) in the presence or absence of cholesterol(5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD) for 24 h and intracellular cholesterol was labeled with lipin staining. Representative confocal images from four indepen-dent experiments are shown. (D) MCF-7 or MDA-MB-231 cells were treated with 10 M tamoxifen (TMX) in the presence or absence of cholesterol (Chol, 5 g/ml)/cyclodextrin (CD, 0.1%) complex (Chol/CD) for 24 h. The cell viability was quantied by AlamarBlue staining and was plotted using the GraphPad Prism 5.0 software. Datarepresent mean standard deviation (SD) from four independent experiments. **P < 0.01 between two indicated groups.

ARTICLE IN PRESS



Q3Q4

12345678

9

10111213141516171819202122232425

26272829303132333435363738394041

Technology Development Fund (FDCT) of Macau SAR (FDCT/119/2013/A3) (to J.S.S and J.O.L) and Matching Research Grant of theUniversity of Macau (MRG002/JSS/2015/FHS) (to J.S.S).

Conict of interest

The authors have declared that no conict of interest exists.

Appendix: Supplementary material

Supplementary data to this article can be found online atdoi:10.1016/j.canlet.2015.03.022.

References

[1] L.J. Lerner, V.C. Jordan, Development of antiestrogens and their use in breastcancer: eighth Cain memorial award lecture, Cancer Res. 50 (1990) 41774189.

[2] V.C. Jordan, Long-term adjuvant tamoxifen therapy for breast-cancer, BreastCancer Res. Treat. 15 (1990) 125136.

[3] V.C. Jordan, C.S. Murphy, Endocrine pharmacology of antiestrogens as antitumoragents, Endocr. Rev. 11 (1990) 578610.

[4] C. Knabbe, G. Zugmaier, M. Schmahl, M. Dietel, M.E. Lippman, R.B. Dickson,Induction of transforming growth-factor-beta by the antiestrogens droloxifene,tamoxifen, and toremifene in MCF-7 cells, Am. J. Clin. Oncol. 14 (1991) S15S20.

[5] K.L. Blackwell, Z.A. Haroon, S. Shan, W. Saito, G. Broadwater, C.S. Greenberg,et al., Tamoxifen inhibits angiogenesis in estrogen receptor-negative animalmodels, Clin. Cancer Res. 6 (2000) 43594364.

[6] A. Gagliardi, D.C. Collins, Inhibition of angiogenesis by antiestrogens, CancerRes. 53 (1993) 533535.

[7] E.F. Haran, A.F. Maretzek, I. Goldberg, A. Horowitz, H. Degani, Tamoxifenenhances cell-death in implantedMCF7 breast-cancer by inhibiting endotheliumgrowth, Cancer Res. 54 (1994) 55115514.

[8] J. Heidemann, H. Ogawa, M.F. Otterson, V.B. Shidham, D.G. Binion, Antiangiogenictreatment of mesenteric desmoid tumors with toremifene and interferonalfa-2b: report of two cases, Dis. Colon Rectum 47 (2004) 118122.

[9] J.A. Hurteau, M.F. Brady, K.M. Darcy, W.P. McGuire, P. Edmonds, M.L. Pearl, et al.,Randomized phase III trial of tamoxifen versus thalidomide in women withbiochemical-recurrent-only epithelial ovarian, fallopian tube or primaryperitoneal carcinoma after a complete response to rst-line platinum/taxanechemotherapy with an evaluation of serum vascular endothelial growth factor(VEGF): a Gynecologic Oncology Group Study, Gynecol. Oncol. 119 (2010)444450.

[10] T. Mele, D. Generali, S. Fox, M.P. Brizzi, A. Bersiga, M. Milani, et al., Anti-angiogenic effect of tamoxifen combined with epirubicin in breast cancerpatients, Breast Cancer Res. Treat. 123 (2010) 795804.

[11] F.R. Maxeld, I. Tabas, Role of cholesterol and lipid organization in disease,Nature 438 (2005) 612621.

[12] J.P. Incardona, S. Eaton, Cholesterol in signal transduction, Curr. Opin. Cell Biol.12 (2000) 193203.

[13] L. Liscum, N.K. Dahl, Intracellular cholesterol transport, J. Lipid Res. 33 (1992)12391254.

[14] E. Ikonen, Cellular cholesterol tracking and compartmentalization, Nat. Rev.Mol. Cell Biol. 9 (2008) 125138.

[15] H.J. Kwon, L. Abi-Mosleh, M.L. Wang, J. Deisenhofer, J.L. Goldstein, M.S. Brown,et al., Structure of N-terminal domain of NPC1 reveals distinct subdomains forbinding and transfer of cholesterol, Cell 137 (2009) 12131224.

[16] K.B. Peake, J.E. Vance, Defective cholesterol tracking in Niemann-PickC-decient cells, FEBS Lett. 584 (2010) 27312739.

[17] J. Xu, Y.J. Dang, Y.R.Z. Ren, J.O. Liu, Cholesterol tracking is required for mTORactivation in endothelial cells, Proc. Natl. Acad. Sci. U.S.A. 107 (2010) 47644769.

[18] B.A. Nacev, P. Grassi, A. Dell, S.M. Haslam, J.O. Liu, The antifungal drugitraconazole inhibits vascular endothelial growth factor receptor 2 (VEGFR2)glycosylation, tracking, and signaling in endothelial cells, J. Biol. Chem. 286(2011) 4404544056.

[19] L.H. Fang, S.H. Choi, J.S. Baek, C. Liu, F. Almazan, F. Ulrich, et al., Control ofangiogenesis by AIBP-mediated cholesterol eux, Nature 498 (2013) 118122.

[20] G. Gimpl, K. Gehrig-Burger, Cholesterol reporter molecules, Biosci. Rep. 27 (2007)335358.

[21] R.A. Dwek, T.D. Butters, F.M. Platt, N. Zitzmann, Targeting glycosylation as atherapeutic approach, Nat. Rev. Drug Discov. 1 (2002) 6575.

[22] V.I. Korolchuk, S. Saiki, M. Lichtenberg, F.H. Siddiqi, E.A. Roberts, S. Imarisio,et al., Lysosomal positioning coordinates cellular nutrient responses, Nat. CellBiol. 13 (2011) 453460.

[23] J.E. Vance, K.B. Peake, Function of the Niemann-Pick type C proteins and theirbypass by cyclodextrin, Curr. Opin. Lipidol. 22 (2011) 204209.

[24] S.R. Goldstein, S. Siddhanti, A.V. Ciaccia, L. Plouffe, A pharmacological reviewof selective oestrogen receptor modulators, Hum. Reprod. Update 6 (2000)212224.

[25] P. Reddy, P. DeFusco, The role of antiestrogens in the treatment and preventionof breast cancer, Formulary 33 (1998) 744768.

[26] T. Takahashi, M. Shibuya, The 230 kDa mature form of KDR/Flk-1 (VEGFreceptor-2) activates the PLC-gamma pathway and partially induces mitoticsignals in NIH3T3 broblasts, Oncogene 14 (1997) 20792089.

[27] Y. Sancak, L. Bar-Peled, R. Zoncu, A.L. Markhard, S. Nada, D.M. Sabatini,Ragulator-Rag complex targets mTORC1 to the lysosomal surface and isnecessary for its activation by amino acids, Cell 141 (2010) 290303.

[28] V.I. Korolchuk, D.C. Rubinsztein, Regulation of autophagy by lysosomalpositioning, Autophagy 7 (2011) 927928.

[29] R. Puertollano, mTOR and lysosome regulation, F1000Prime Rep. 6 (2014) 52.[30] Y. Chen, M. Schindler, S.M. Simon, A mechanism for tamoxifen-mediated

inhibition of acidication, J. Biol. Chem. 274 (1999) 1836418373.[31] R. Usui, M. Shibuya, S. Ishibashi, Y. Maru, Ligand-independent activation of

vascular endothelial growth factor receptor 1 by low-density lipoprotein, EMBORep. 8 (2007) 11551161.

[32] F.Y. Jin, N. Hagemann, U. Brockmeier, S.T. Schafer, A. Zechariah, D.M. Hermann,LDL attenuates VEGF-induced angiogenesis via mechanisms involving VEGFR2internalization and degradation following endosome-trans-Golgi networktracking, Angiogenesis 16 (2013) 625637.

[33] E.S. Antonarakis, E.I. Heath, D.C. Smith, D. Rathkopf, A.L. Blackford, D.C. Danila,et al., Repurposing itraconazole as a treatment for advanced prostate cancer:a noncomparative randomized phase II trial in men with metastatic castration-resistant prostate cancer, Oncologist 18 (2013) 163173.

Fig. 8. Effects of SERM and cholesterol on HUVEC tube formation. (A) HUVEC were seeded onto a Matrigel-coated plate to promote tube formation. Cells were treated with5 M tamoxifen (TMX) or 5 M toremifene (TRM) in the presence or absence (NT) of cholesterol (5 g/ml)/cyclodextrin (0.1%) complex (Chol/CD) for 18 h. The tube forma-tion was visualized by staining with Calcein-AM under a uorescence microscope. Representative uorescence images from six independent experiments are shown.(B) Total tube lengths, sizes and number of junctions from the uorescence images from six experiments were quantied using the AngioQuant software. **P < 0.01 betweentwo indicated groups.

ARTICLE IN PRESS



12345

6

789

10111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455

5657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109

[34] C.M. Rudin, J.R. Brahmer, R.A. Juergens, C.L. Hann, D.S. Ettinger, R. Sebree, et al.,Phase 2 study of pemetrexed and itraconazole as second-line therapy formetastatic nonsquamous non-small-cell lung cancer, J. Thorac. Oncol. 8 (2013)619623.

[35] D.J. Kim, J. Kim, K. Spaunhurst, J. Montoya, R. Khodosh, K. Chandra, et al.,Open-label, exploratory phase II trial of oral itraconazole for the treatment ofbasal cell carcinoma, J. Clin. Oncol. 32 (2014) 745751.

ARTICLE IN PRESS



1234

567

Inhibition of angiogenesis by selective estrogen receptor modulators through blockade of cholesterol trafficking rather than estrogen receptor antagonism Introduction Materials and methods Cells and reagents Filipin staining Immunofluorescence imaging [3H]-thymidine DNA incorporation assay AlamarBlue cell viability assay Western blot analysis Endothelial cell tube formation assay Statistical analysis Results Tamoxifen and SERM induce NPC-like phenotype in HUVEC Tamoxifen traps VEGFR2 in Golgi and induces perinuclear positioning of lysosomes in HUVEC Tamoxifen inhibits VEGFR2 glycosylation and mTOR activity in HUVEC Cholesterol reverses the inhibitory effects of tamoxifen on VEGFR2 and mTOR signaling in HUVEC Cholesterol reverses the inhibitory effects of tamoxifen on HUVEC proliferation Inhibition of cholesterol trafficking by tamoxifen is independent of ER Cholesterol reverses the inhibitory effect of tamoxifen on HUVEC tube formation in Matrigel Discussion Acknowledgements Conflict of interest Supplementary material References

inhibition of angiogenesis by selective estrogen receptor modulators 2015

Documents

mtor activitiesby tamoxifen

inhibition of vegfr2

ernegative breast cancers

underlying mechanism

mechanism of action

sarah head b

ruojing li b

underlying molecularmechanism