ology 81 (2006) 108–114www.elsevier.com/locate/yexmp

Experimental and Molecular Path

Reduction of angiocidin expression in human umbilical vein endothelial cellsvia siRNA silencing inhibits angiogenesis

Xiao Yang, Vicki L. Rothman, Darryl Z. L'Heureux, George Tuszynski ⁎

Department of Neuroscience, Center for Neurovirology, Temple University School of Medicine, 1900 North 12th Street, Philadelphia, PA 19122, USA

Received 9 June 2006Available online 10 August 2006

Abstract

Angiocidin, a protein over-expressed in many different solid tumors and tumor capillary endothelial cells inhibits angiogenesis and tumorgrowth [Zhou, J., et al., 2004. Cloning and characterization of angiocidin, a tumor cell binding protein for thrombospondin-1. J Cell Biochem. 92,125–146]. Since several splice variants of angiocidin have distinct biochemical functions in membrane transport and protein degradation, wesought to evaluate the function of endogenously expressed angiocidin in human umbilical vein endothelial (HUVE) cells using siRNA. Weobserved a 90% reduction of the target mRNA levels after 24 h. Endogenous angiocidin protein expression was reduced by 80% after three days,as evaluated by Western blot analysis. We also found that anti-angiocidin siRNA down-regulated 90% of the protein expression of matrixmetalloproteinase 2 (MMP-2) and 50% of its gelatinolytic activity. Reduction of endogenous angiocidin completely inhibited endothelial cordformation on Matrigel. Cells expressing low levels of angiocidin grew more slowly, were less invasive and less adhesive than control cells.Consistent with the reported function of one of the angiocidin analogues S5a, we found that the expression of polyubiquitinated proteins washigher in anti-angiocidin siRNA-treated cells as compared to normal and control siRNA-treated cells. These results suggest that endogenousangiocidin and its homologues promote endothelial cell invasion, adhesion, and angiogenesis through mechanisms involving polyubiquitin-dependent protein degradation and MMP-2 expression.© 2006 Elsevier Inc. All rights reserved.

Keywords: Angiogenesis; Angiocidin; siRNA; Tumor progression

Introduction

Tumor angiogenesis is a complex process mediated bynumerous inhibitors and activators (Folkman, 2006). Many ofthese molecules have been identified as targets for the treatmentof cancer. These include such molecules as vascular endothelialgrowth factor (VEGF), proteolytic fragments of host moleculessuch as plasminogen and collagen, and integrin adhesionreceptors.

Our laboratory identified a protein from lung tumor extractsthat bound thrombospondin-1 (TSP-1) (Tuszynski et al., 1993).TSP-1 is a large secreted glycoprotein that modulatesangiogenesis (Sargiannidou et al., 2004) and the TSP-1 bindingprotein was postulated to play a role in TSP-1-mediatedmechanisms of angiogenesis. The protein showed high expres-

⁎ Corresponding author.E-mail address: gpt@temple.edu (G. Tuszynski).

0014-4800/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.yexmp.2006.06.003

sion in tumor tissue (Arnoletti et al., 1994; Tuszynski andNicosia, 1994), tumor-associated microvessels (Arnoletti et al.,1994), and tumor stroma (Roth et al., 1997). We now havecloned this TSP-1 binding protein and discovered that theexpressed recombinant protein displayed potent anti-tumoractivity as well as anti-angiogenic activity in a mouse Lewislung model of tumor growth (Zhou et al., 2004). Based on theinhibitory activity of the recombinant protein in tumorprogression as well as its inhibitory activity in angiogenesis,we have named the protein angiocidin.

The sequence of angiocidin shows a high degree ofhomology to two previously described proteins S5a (Deverauxet al., 1995) and antisecretory factor (Johansson et al., 1995)differing from these proteins by only three amino acids in itscarboxyl terminal region. Antibodies prepared against recom-binant angiocidin recognize a single polypeptide chain having amolecular weight of 55 kDa from extracts of endothelial andtumor cells suggesting that our antibody cross-reacts with all the

109X. Yang et al. / Experimental and Molecular Pathology 81 (2006) 108–114

isoforms of angiocidin. Although angiocidin and its homo-logues are similar in structure, they have been reported to havediverse functions. The protein S5a is localized in the 26Sproteosome subunit and is a major protein polyubiquitinrecognition domain. In contrast, antisecretory factor, a proteinwith a reported sequence identical to that of S5a, is a secretedprotein that functions in the regulation of intestinal fluidtransport induced by cholera toxin (Johansson et al., 1995).

As mentioned above, angiocidin differs from S5a andantisecretory factor by three additional amino acids in itscarboxyl terminal domain (Zhou et al., 2004). Angiocidinshares many of the binding motifs present in S5a andantisecretory factor. Our laboratory has investigated thesignificance of some of these peptide domains as well asother regions of the molecule in mediating the anti-angiogenicactivity of angiocidin. We found that a seven amino acidsequence that has been reported to mediate the antisecretoryactivity of antisecretory factor (Johansson et al., 1997b) playedno role in the angiogenic activity of angiocidin while one of twopolyubiquitin binding sequences present near the carboxylterminus was crucial for the anti-angiogenic activity ofangiocidin (Dimitrov et al., 2005). Additionally, we haverecently identified a 20 amino acid domain in the aminoterminal domain of angiocidin that also mediates the anti-tumorand anti-angiogenic activities of angiocidin (Zhou et al., 2004).This domain is crucial for the interaction of angiocidin with theextracellular matrix and the integrin adhesion receptor, α2β1(Sabherwal et al., in press). Synthetic peptides corresponding tothis sequence promoted cell adhesion and inhibited tumorgrowth, mimicking the anti-tumor activity of angiocidin(Sabherwal et al., in press).

A major focus of our research is to discover how angiocidinblocks tumor progression and angiogenesis despite the fact thattumor cells and tumor vasculature over-express angiocidin andits homologues. If angiocidin is a tumor suppressor protein, thenwhy do tumor cells over-express this protein and itshomologues and why should our recombinant protein inhibittumor growth? One explanation is that angiocidin and itshomologues function as endogenous inhibitors of angiogenesisand tumor growth in much the same way as has been shown forthe tumor inhibitor angiostatin by Folkman and colleagues(O'Reilly et al., 1997) as well as a number of other circulatinginhibitors produced as a consequence of growing tumors such asendostatin, tumstatin, and arrestin (Folkman, 2006). Theseinhibitors circulate and bind to tumors at distant sites and inhibittheir growth and vascularization. Consistent with this concept,we have detected nanogram–microgram per milliliter levels ofangiocidin in the sera of patients with prostate, breast, andhepatocellular carcinoma while the sera of healthy individualscontained levels undetectable by our assay system, which candetect low picogram per milliliter levels of angiocidin(Sabherwal et al., 2004, 2005a,b).We therefore hypothesizethat angiocidin is a secreted tumor suppressor protein andfunctions externally by modulating matrix cellular adhesion andangiogenesis.

In this study, we provide support for this hypothesis byshowing that endothelial cells that are inhibited from expressing

angiocidin exhibit a nonangiogenic phenotype and that criticalenzymes, such as MMP-2 that remodels the extracellularmatrix, are regulated by endogenous angiocidin. Our studies arethe first to provide direct evidence that angiocidin and itsanalogues are required for angiogenesis and that endogenousangiocidin plays a role in cell matrix interactions.

Materials and methods

Antibodies and reagents

All chemicals were reagent grade and unless specified otherwise werepurchased from Fisher Scientific, Pittsburgh, PA. Monoclonal and polyclonalantibodies against angiocidin were prepared from purified recombinantprotein (Covance, Denver PA). Goat anti-rabbit IgG-Texas red-conjugatedantibody and fluorescein Isothiocyanate (FITC)-conjugated anti-mouse IgGwere obtained from Vector Laboratories, Burlingame, CA. Rabbit anti-ubiquitin antibody was purchased from Biosource, Camarillo, CA. Thebicinchoninic acid (BCA) protein assay kit was purchased from PierceChemical Co, Rockford, IL. Small interference RNA (siRNA), transfectionreagents, reverse transcription polymerase chain reaction (RT-PCR) reagentswere purchased from Qiagen, Valencia, CA. DNase treatment and RemovalReagent were purchased from Ambion, Austin, TX. Mouse anti-humanMMP-2 antibody was purchased from R and D Systems, Inc., Minneapolis,MN. Trizol reagent was purchased from Invitrogen Life Technologies,Carlsbad, California.

Cell culture

HUVE cells were grown in EBM-2 medium (Cambrex Corporation, EastRutherford, NJ) supplemented with the EBM-2 bullet kit, which contains 2%fetal bovine serum (FBS), and cultured in 5% CO2 at 37°C. HUVE cells wereused before the 10th passage, because after 15 passages the endothelial cellsundergo morphological and functional changes that make them unsuitable forangiogenesis assessment.

siRNA treatment

Specific siRNA directly against angiocidin was designed by Qiagen Inc. Thesense sequence is GCAGGAUGCUGUCAACAUA and the antisense sequenceis UAUGUUGACAGCAUCCUGC, which targeted the sequence CAGCAG-GATGCTG TCAACATA in angiocidin. A nonsilencing siRNAwas used as thecontrol. The sense sequence for the control is UUCUCCGAACGUGUCAC-GUdTdT, where dTdT are dinucleotide overhangs, and its antisense sequence isACGUGACACGUUCGGAGAAdTdT. The siRNAs were introduced intoHUVE cells by lipid-mediated transfection. For the transfection experiments,4.0×105 cells per well were seeded in a 6-well plate and incubated overnight innormal growth medium to reach 80–90% confluency. The culture medium wasaspirated and cells were transfected with 100 nM siRNA (either specific toangiocidin siRNA or nonsilencing siRNA) in the presence of 7.5 μl RNAifectTransfection Reagent in 2 ml of normal growth media.

RNA extraction and RT-PCR

After 24 h of transfection, total RNAwas extracted using the Trizol reagentaccording to the manufacturer's recommendation. The RNA sample was treatedwith DNase I to remove the contaminating DNA. This purified RNA wasreversely transcribed to single-stranded cDNAs using a oligodeoxythymidylicacid primer with reverse transcriptase. PCR amplification using angiocidinspecific primers was used to monitor angiocidin expression in untreated cells,siRNA (against angiocidin)-treated cells, and nonsilencing siRNA-treated cells.Angiocidin primers were chosen in order to amplify the full-length angiocidincDNA sequence. The primer A (sense) is 5′ATGGTGTTGGAAAGCACTATGGTG 3′ and the primer B (antisense) is 5′ CTT CTTGTCT T CCTCCTTCTTGTC 3′. β-actin expression was also monitored in untreated cells. Samplesfor analysis consisted of RNA from anti-angiocidin siRNA and nonsilencing

Fig. 1. Anti-angiocidin siRNA down-regulates expression of angiocidin inHUVE cells: HUVE cells were either untreated or treated with anti-angiocidinsiRNA or control siRNA as described in Materials and methods. Panel A showsangiocidin and β-actin cDNA generated by RT-PCR from HUVE cells treatedfor 24 h with the following: Lane 1, buffer-treated cells; lane 2, cells treated withfirst anti-angiocidin siRNA construct; lane 3, cells treated with second anti-angiocidin siRNA construct; lane 4, cells treated with control siRNA construct.Panel B shows anti-angiocidin antibody and anti-β-actin antibody Western blotof HUVE cells treated for 3 days with either buffer (lane 1), control siRNA (lane2), or the first anti-angiocidin siRNA construct (lane 3).

110 X. Yang et al. / Experimental and Molecular Pathology 81 (2006) 108–114

siRNA-treated cells as well as buffer-treated cells. These reactions involvedinitial denaturation at 94°C for 3 min followed by 38 cycles at 94°C(denaturation) for 30 s, 65°C (annealing) for 1 min, and 68°C (extension) for 2min. The PCR products were analyzed on a 1.5% agarose gel which contained2% ethidium bromide and photographed.

Evaluation of protein expression by Western blotting

After 3 days of transfection, protein was extracted from untreated cells,control siRNA-treated cells, and anti-angiocidin siRNA-treated cells with theBiosource cell extraction buffer containing protease inhibitors. Proteinconcentrations were determined using the BCA protein assay. Equal amountsof protein from each treatment group (10 μg) were separated by SDS-PAGE andtransferred to polyvinylidene difluoride membranes. The membranes wereincubated in blocking buffer consisting of Tris-buffered saline (TBS) containing5% nonfat dry milk for 1 h. The membranes were then incubated with a 1:2000dilution of rabbit polyclonal anti-angiocidin (initial concentration: 1.6 mg/ml)over night at 4°C or incubated with a 1:1000 dilution of rabbit poly-ubiquitinantiserum over night at 4°C. The next day, the membranes were incubated with a1:10,000 dilution of horseradish peroxidase-conjugated goat anti-rabbitsecondary antibody (initial concentration 2 mg/ml) for 1 h at room temperature.After incubation with the ECL reagent, chemiluminescence signals wererecorded on X-ray film and quantitated by image analysis. Membranes werethen stripped and probed with anti-β-actin antibody to determine protein loadingas previously described (Zhou et al., 2004).

Zymography

The MMP expression in untreated cells, siRNA (against angiocidin) ornonsilencing siRNA-treated cells was evaluated by zymography as previouslydescribed (Qian et al., 1997). Three days after transfection, total cellular proteinwas extracted with lysis buffer (1% NP-40 in 50 mM Tris–HCl buffer, pH 8.0containing 150 mM NaCl). Protein concentration was determined using theBCA protein assay. Equal amounts of protein from the treatment groups wererun under nonreducing conditions. Gels were run at 20 mA for 4–5 hours.Following electrophoresis, the SDS was removed from the gel by incubation inthe washing buffer (50 mM Tris–HCl, pH 7.5, 2.5% Triton X-100). Then the gelwas incubated in developing buffer (50 mM Tris–HCl, pH 7.5, containing 5 mMCaCl2, 150 mM NaCl, 0.02% NaN3) overnight at 37°C. The zymogram wasstained with 0.05% Coomassie Blue. Areas of digestion appear as clear bandsagainst a darkly stained background where the substrate has been degraded bythe enzyme.

Capillary tube formation on Matrigel matrix

The capacity of transfected HUVE cells to form capillary networks wasevaluated using the Matrigel angiogenesis assay. Briefly, 20,000 HUVE cellswere plated in 96 well-microtiter plates coated with Matrigel as previouslydescribed (Grant et al., 1992). Control and cells transfected with anti-angiocidin siRNA or control siRNA were plated in complete media and tubeformation was evaluated after 24 h by viewing the wells with phase contrastmicroscopy at 100×.

Cell invasion assay

Cell invasion of control and siRNA-treated cells was performed aspreviously described (Zhou et al., 2004). Briefly, 50,000 cells were plated inserum-free media on 8 μm diameter pore top insets coated with type I collagen.Bottom chambers contained complete media. After 3 h, the cells on the bottomof the top insert were fixed with 2% glutaraldehyde, stained, and the totalnumber of cells counted.

Cell adhesion assay

Cell adhesion of control and siRNA-treated cells was performed in 96-wellplates coated with either collagen I or bovine serum albumin (BSA) as

previously described (Zhou et al., 2004). After 1 h, cells were fixed, stained, andthe total number of cells per 4 representative fields was counted under lightmicroscopy at 100×.

Immunocytochemistry

HUVE cells, growing in two chamber slides, were transfected withanti-angiocidin siRNA or control siRNA. After 4 days, they were fixedwith 4% paraformaldehyde for 30 min at room temperature. Cells werewashed with 0.2% Triton X-100 in PBS for 90 s. Then cells were washedthree times with PBS and blocked with 5% horse serum in 0.1% BSA inPBS for 1 h. Cells were then incubated with 1:250 dilution of rabbit anti-ubiquitin serum for 1 h. Cells were washed three times with PBS, thenincubated with a 1:1000 dilution of Texas red labeled goat anti-rabbit IgGfor 45 min. Cells were then washed three times with PBS and incubatedwith 2 μg/ml anti-angiocidin monoclonal antibody for 1 h. Cells werewashed with PBS and incubated with FITC-conjugated anti-mouse IgG for45 min in the dark. After washing, slides were mounted and viewed byfluorescence microscopy at 400×. Controls in which primary antibody wasomitted were negative.

Results

Reduction of angiocidin expression in HUVE cells by siRNA

HUVE cells were transfected with angiocidin-targetedsiRNA and nonsilencing control siRNA. After 24 hours oftransfection with 100 nM angiocidin siRNA, angiocidinmRNA expression was down regulated to 90% compared tountreated cells (Fig. 1A). The control siRNA had nosignificant effect on angiocidin mRNA expression inHUVE cells. The cell lysates were analyzed by WesternBlot analysis for angiocidin expression after 72 h oftransfection. The expression of angiocidin was down-regulated to 80% as compared to control siRNA or untreatedcells (Fig. 1B). Expression levels of the housekeepingprotein β-actin were unchanged in the various treatmentgroups and therefore this protein was considered a loadingcontrol in the Western blot experiments. These resultsindicate that anti-angiocidin siRNA effectively and specifi-cally down-regulated angiocidin mRNA and protein expres-sion in HUVE cells.

Fig. 3. Anti-angiocidin siRNA inhibits adhesion and invasion of HUVE cells:HUVE cells either untreated or treated with anti-angiocidin siRNA or controlsiRNA for 3 days and evaluated for adhesion (A) or invasion of collagen (B) asdescribed in Materials and methods. The error bars represent the standard errorof the mean of triplicate samples and the experiment was repeated twice withsimilar results.

111X. Yang et al. / Experimental and Molecular Pathology 81 (2006) 108–114

Down-regulation of angiocidin in HUVE cells inhibits tubeformation on Matrigel

To evaluate the impact of angiocidin knockdown on HUVEcell tube formation, we performed a Matrigel matrix tubeformation assay using the anti-angiocidin siRNA transfectedHUVE cells as well as untreated and control siRNA transfectedcells. When compared with untreated cells or cells transfectedwith nonsilencing control siRNA, anti-angiocidin siRNAtransfected cells lost the ability to form tubular networks onMatrigel. The cells treated with anti-angiocidin siRNAaggregated, but failed to form tube-like networks (Fig. 2),suggesting that HUVE cell tube formation on Matrigel isregulated by the expression of angiocidin. These results indicatethat endogenous angiocidin is required for HUVE cell tubeformation.

Down-regulation of angiocidin in HUVE cells inhibitsadhesion to type I collagen

To access the potential effect of endogenous angiocidinexpression on cell adhesion, we performed a collagen-endothelial cell adhesion assay. After a 72 h transfection withanti-angiocidin siRNA or control siRNA, the level of adhesionto type I collagen in anti-angiocidin siRNA transfected cells wassignificantly lower than that observed in control siRNAtransfected HUVE cells or normal cells. We found that cellswith decreased angiocidin expression had almost a 70%reduction in the ability to adhere to type I collagen (Fig. 3A).These results indicate that in the absence of sufficient levels ofangiocidin adhesion to type I collagen is impaired and isconsistent with the inability of angiocidin-null cells to formmicrovascular networks on Matrigel, a process that involvesadhesive interactions to matrix components such as collagen(Sweeney et al., 2003).

Down-regulation of angiocidin expression reduces HUVEcells’ capability to invade type I collagen

To assess the effect of endogenous angiocidin on HUVEcells invasion of collagen, we performed migration assays usingcollagen I-coated transwell Boyden chambers and serum as thechemo-attractant. As observed in our adhesion assays, we found

Fig. 2. Anti-angiocidin siRNA treatment inhibits HUVE cell tube formation on MatrigsiRNA for 3 days and were evaluated for tube formation on Matrigel as described in Mmagnification of 100×.

that the invasive capacity of the cells transfected with anti-angiocidin siRNA was significantly decreased compared withthe invasive activity of the cells transfected with control siRNAor treated with buffer (Fig. 3B). Invasion of the anti-angiocidinsiRNA transfected HUVE cells was reduced to 60% of that ofthe control cells (normal cells or cells treated with nonsilencingsiRNA). These experiments demonstrate that endogenousangiocidin is required for HUVE cell invasion.

Down-regulation of angiocidin expression reduces MMP-2activity in HUVE cells

In order to characterize the effects of angiocidin expressionon MMP-2, we used zymography to detect the activity of

el: HUVE cells either untreated or treated with anti-angiocidin siRNA or controlaterials and methods. Tubes were visualized by phase contrast microscopy at a

Fig. 4. Anti-angiocidin siRNA treatment inhibits expression and activity ofMMP-2 in HUVE cells: HUVE cells were either treated with anti-angiocidinsiRNA or control siRNA for 3 days and cell lysates were evaluated for MMP-2activity by zymography or protein expression by Western blot analysis asdescribed in Materials and methods. Loading controls were performed with anti-β-actin antibody.

Fig. 5. Anti-angiocidin siRNA treatment promotes expression of ubiquitinatedproteins in HUVE cells: HUVE cells either untreated or treated with anti-angiocidin siRNA or control siRNA for 3 days and cell lysates were evaluatedfor ubiquitinated protein expression by Western blot analysis as described inMaterials and methods. Loading controls were performed with anti-β-actinantibody.

112 X. Yang et al. / Experimental and Molecular Pathology 81 (2006) 108–114

gelatinases in nonsilencing control siRNA-treated HUVE cellsand anti-angiocidin siRNA-treated HUVE cells. The activity ofMMP-2 (72 kDa gelatinase) in anti-angiocidin siRNA-treatedHUVE cells was significantly reduced compared to that ofcontrol siRNA-treated cells (Fig. 4). We also did Western blotanalysis of whole cell lysates to compare MMP-2 proteinexpression in control siRNA-treated HUVE cells and anti-angiocidin siRNA-treated HUVE cells (Fig. 4). Our results wereconsistent with the results obtained with zymography, showinga more than 70% reduction in MMP-2 protein expression ascompared to the control cells. These results indicate thatdecreased activity of MMP-2 in angiocidin null cells is due to aninhibition of protein expression. Therefore, endogenousangiocidin appears to be an important factor in the regulationof MMP-2 expression. Since MMP-2 plays an important role inHUVE cell invasion (Taraboletti et al., 2000), these results areconsistent with our results showing that angiocidin null cells areless invasive than control cells (Fig. 3B).

Down-regulation of angiocidin increases the expression ofubiquinated protein in HUVE cells

In a previous study, we showed that angiocidin possessedat least two polyubiquitin binding sites and that at least onedomain functioned in promoting the apoptotic activity ofexogenously-added angiocidin (Dimitrov et al., 2005).Additionally, angiocidin shares these same polyubiquitinbinding domains with an internal proteasome subunit, S5a,that functions to bind and facilitate the degradation ofpolyubiquitinated proteins (Deveraux et al., 1995) inendothelial cells. Therefore, due to the close identity betweenS5a and angiocidin and the fact that our angiocidin siRNAwould be expected to silence both proteins, we sought toevaluate the effect angiocidin siRNA on the expression ofubiquitinated proteins. We used a rabbit anti-ubiquitin anti-body to evaluate ubiquinated protein expression by Westernblot analysis. We found that the total level of ubiquinatedproteins was higher in anti-angiocidin siRNA-treated HUVEcells compared to the buffer-treated HUVE cells or cellstreated with nonsilencing siRNA (Fig. 5). To confirm ourWestern blot results, we immunostained anti-angiocidinsiRNA-treated HUVE cells, buffer-treated HUVE cells, andnonsilencing siRNA-treated HUVE cells with a mouse

monoclonal anti-angiocidin antibody and rabbit anti-ubiquitinantibody (Fig. 6). Immunostaining with an anti-angiocidinantibody showed that the level of angiocidin expression wassignificantly lower in anti-angiocidin siRNA-treated cellscompared with the control cells. In contrast, immunostainingwith an anti-ubiquitin antibody showed that the level ofubiquitinated proteins was relatively higher in anti-angiocidinsiRNA-treated cells as compared to controls. Both angiocidinand ubiquitin co-localized in all the groups, consistent withour previous observations showing that angiocidin bindsubiquitinated proteins (Dimitrov et al., 2005) (Fig. 6).Therefore, down-regulation of angiocidin correlates with anincrease expression of ubiquitinated proteins in HUVE cells.

Discussion

Our laboratory previously isolated a protein from lung tumorextracts by peptide affinity chromatography using the type 1TSP-1 repeat peptide CSVTCG. A single protein peak wasisolated from tumor cells which also analyzed as a single bandon SDS gels under nonreducing conditions (Tuszynski et al.,1993). The protein was cloned from a prostate cell cDNAlibrary and the full-length cDNA was nearly identical to thatreported for S5a, a polyubiquitin recognition subunit (Deverauxet al., 1995) and antisecretory factor (Johansson et al., 1997a), asecreted protein that inhibits intestinal water secretion inresponse to cholera toxin. We have named our cloned proteinangiocidin due to its in vitro anti-angiogenic activity and in vivoanti-tumor activity (Zhou et al., 2004). In vitro, recombinantangiocidin induced apoptosis in both endothelial and tumorcells grown in tissue culture, as well as inhibiting cell invasionand endothelial cell tube formation. In vivo, angiocidin showeda potent anti-tumor activity at the relatively low dose of 0.4 mg/kg administered intravenously to mice bearing Lewis lungcarcinoma.

In an effort to understand how angiocidin exerts its anti-tumor activity, we decided to evaluate the function ofendogenous HUVE cell angiocidin in the adhesive interactions

Fig. 6. Immunolocalization of angiocidin and ubiquitin in HUVE cells either untreated or treated with anti-angiocidin siRNA or control siRNA for 3 days: Cells werestained for angiocidin and ubiquitin as described in Materials and methods. Cells were photographed by fluorescence microscopy at a magnification of 200×.

113X. Yang et al. / Experimental and Molecular Pathology 81 (2006) 108–114

important in angiogenesis. In this study, we use anti-angiocidinsiRNA to inhibit the expression of angiocidin in HUVE cells.The relatively new technique of RNA interference with shortdouble stranded RNA molecules that are complementary to thetarget mRNA is a potentially powerful tool to silence genes andevaluate their function (Agrawal et al., 2003).

Our studies show that the expression of endogenousangiocidin in HUVE cells can be silenced with siRNA againstangiocidin. We used two different siRNA molecules preparedagainst angiocidin and demonstrate that 24 h after theseoligonucleotides were transfected into HUVE cells, endogenousangiocidin mRNA was reduced by almost 80% (Fig. 1). Incontrast, nonsilencing RNA molecules had no effect onangiocidin message indicating that the transfection procedurehad no effect on gene expression. After 3 days of treatment withour specific angiocidin siRNA, angiocidin protein expressionwas decreased significantly by more than 70%. In ourexperiments, we used short 19 nucleotides double-strandRNA molecules. These short siRNA molecules at the relativelylow doses used in our study would not be expected to up-regulate interferon-induced stress genes and their pathways(Elbashir et al., 2001). Since the structure of angiocidin ishighly homologues to S5a and antisecretory, our angiocidinsiRNA preparations would be expected to down-regulate thesemolecules as well.

The functional consequences of silencing angiocidinexpression in HUVE cells were investigated using adhesion,invasion, and Matrigel tube formation. In all three assays,HUVE cells with down-regulated angiocidin showed inhibitionof activity as compared to nonsilencing siRNA-treated cellsand buffer-treated cells. For example, HUVE cells withdiminished angiocidin expression were less invasive andadhesive as compared to control cells (Fig. 3) and failed todevelop vascular networks on Matrigel (Fig. 2). These resultswere somewhat surprising in view of our previous resultsshowing that exogenously added angiocidin inhibited HUVEcell invasion, adhesion, and tube formation on Matrigel (Zhouet al., 2004). We postulate that angiocidin can function as a

secreted endogenous inhibitor of angiogenesis, yet inhibitingexpression of endogenous angiocidin produces a less angio-genic phenotype. One possible explanation for these resultsmay be that two populations of angiocidin exist. A secretedpopulation binds collagen and inhibits its adhesive propertiesand a membrane-associated population binds adhesion recep-tors and acts an adhesion co-receptor promoting cellularadhesion. In support of this hypothesis, we have recentlyshown that angiocidin binds collagen with high affinity andinhibits its cell adhesive activity as well as binding α2β1integrin receptors (Sabherwal et al., in press). Binding ofangiocidin to cell surface α2β1 integrins activates the integrinreceptor promoting adhesion. In support of this explanation,we have shown that angiocidin activates focal adhesion kinaseand activates other integrin-associated signaling moleculessuch as p38 (unpublished data). Therefore, when angiocidin issilenced, integrin receptors are less active and cells assume anonadhesive and nonangiogenic phenotype.

Additionally, we have discovered that angiocidin silencedcells secrete and express less MMP-2 than control cells (Fig. 4).This is a new finding and suggests that angiocidin may regulateMMP-2 expression. The observation that cells with decreasedangiocidin expression produce less MMP-2 is consistent withtheir diminished invasive activity, since MMP-2 and relatedmatrix remodeling enzymes are required for efficient invasionof collagen (Hornebeck et al., 2002).

Finally, we made the intriguing observation that cells treatedwith siRNA against angiocidin express higher levels ofubiquitinated proteins as compared to controls (Figs. 5 and 6).Since angiocidin is nearly analogous to the polyubiquitinatedprotein recognition subunit of the proteasome, S5a, we wouldexpect that S5a as well as angiocidin was down-regulated in ourangiocidin siRNA-treated HUVE cells. Therefore, we wouldexpect that degradation of ubiquitinated proteins would beslower in the angiocidin siRNA-treated cells than in controlcells resulting in a greater accumulation of ubiqitinated proteinsin the angiocidin siRNA-treated cells as compared to the controlcells.

114 X. Yang et al. / Experimental and Molecular Pathology 81 (2006) 108–114

In summary, our data show that anti-angiocidin siRNAreduced the expression of endogenous angiocidin in HUVEcells. Silencing the expression of endogenous angiocidininhibited tube formation, cell adhesion, and cell invasion. Ourdata also show that reducing the expression of angiocidininhibits the activity and expression of matrix metalloproteinase2 (MMP-2), a key enzyme that mediates the invasive andangiogenic activity of endothelial cells. Finally, our resultsshow that down-regulation of angiocidin and presumably itsanalogues likely inhibited protein degradation through theubiquitin proteasome pathway. Therefore, angiocidin and itsanalogues are important proteins that regulate angiogenesisthrough a variety of novel mechanisms that include adhesion,invasion, and protein degradation. Our results further suggestthat anti-angiogenic peptides and mimetics based on thestructure of angiocidin should hold promise for the treatmentof cancer.

Acknowledgments

This study was supported in part by a grant from the NationalInstitute of Health R01 CA 88931 to GPT.

References

Agrawal, N., et al., 2003. RNA interference: biology, mechanism, andapplications. Microbiol. Mol. Biol. Rev. 67, 657–685.

Arnoletti, J.P., et al., 1994. Computer-assisted image analysis of tumor sectionsfor a new thrombospondin receptor. Am. J. Surg. 168, 433–436.

Deveraux, Q., et al., 1995. Inhibition of ubiquitin-mediated proteolysis by theArabidopsis 26 S protease subunit S5a. J. Biol. Chem. 270, 29660–29663.

Dimitrov, S., et al., 2005. The endothelial apoptotic activity of angiocidin isdependent on its polyubiquitin binding activity. Br. J. Cancer 93, 662–669.

Elbashir, S.M., et al., 2001. Duplexes of 21-nucleotide RNAs mediate RNAinterference in cultured mammalian cells. Nature 411, 494–498.

Folkman, J., 2006. Angiogenesis. Annu. Rev. Med. 57, 1–18.Grant, D., et al., 1992. Extracellular matrix–cell interaction: matrigel and

complex cellular pattern formation. Lab. Invest. 67, 805–806 (author reply807-8).

Hornebeck, W., et al., 2002. Matrix-directed regulation of pericellularproteolysis and tumor progression. Semin. Cancer Biol. 12, 231–241.

Johansson, E., et al., 1995. Molecular cloning and expression of a pituitary glandprotein modulating intestinal fluid secretion. J. Biol. Chem. 270,20615–20620.

Johansson, E., et al., 1997a. Antisecretory factor suppresses intestinalinflammation and hypersecretion. Gut 41, 642–645.

Johansson, E., et al., 1997b. Identification of an active site in the antisecretoryfactor protein. Biochim. Biophys. Acta 1362, 177–182.

Qian, X., et al., 1997. Thrombospondin-1 modulates angiogenesis in vitro byup-regulation of matrix metalloproteinase-9 in endothelial cells. Exp. CellRes. 235, 403–412.

O'Reilly, M.S., et al., 1997. Endostatin: an endogenous inhibitor of angiogenesisand tumor growth. Cell 88, 277–285.

Roth, J.J., et al., 1997. Histopathology and clinical assessment correlate withthe cysteine–serine–valine–threonine–cysteine–glycine (CSVTCG) recep-tor of thrombospondin-1 in breast tumors. Histol. Histopathol. 12,1013–1018.

Sabherwal, Y., et al., 2004. Angiocidin, an endogenous inhibitor of angiogen-esis, is increased in the sera of cancer patients. Proc. Am. Assoc. Cancer Res.45, 1143.

Sabherwal, Y., et al., 2005a. Clinical significance of serum angiocidin levels inhepatocellular carcinoma. Proc. Am. Assoc. Cancer Res. 46, 1319.

Sabherwal, Y., et al., 2005b. Sensitive capture ELISA for measuring angiocidinin human serum: serum angiocidin levels in healthy volunteers and inprostate cancer patients. Proc. Am. Assoc. Cancer Res. 46, 743.

Sabherwal, Y., et al., in press. Integrin alpha2beta1 mediates the anti-angiogenicand anti-tumor activities of angiocidin, a novel tumor-associated protein.Experimental Cell Research.

Sargiannidou, I., et al., 2004. Mechanisms of thrombospondin-1-mediatedmetastasis and angiogenesis. Semin. Thromb. Hemost. 30, 127–136.

Sweeney, S.M., et al., 2003. Angiogenesis in collagen I requires alpha2beta1ligation of a GFP*GER sequence and possibly p38 MAPK activation andfocal adhesion disassembly. J. Biol. Chem. 278, 30516–30524.

Taraboletti, G., et al., 2000. The heparin binding 25 kDa fragment ofthrombospondin-1 promotes angiogenesis and modulates gelatinase andTIMP-2 production in endothelial cells. FASEB J. 14, 1674–1676.

Tuszynski, G.P., et al., 1993. Identification and characterization of a tumor cellreceptor for CSVTCG, a thrombospondin adhesive domain. J. Cell Biol.120, 513–521.

Tuszynski, G.P., Nicosia, R.F., 1994. Localization of thrombospondin and itscysteine–serine–valine–threonine–cysteine–glycine-specific receptor inhuman breast carcinoma. Lab. Invest. 70, 228–233.

Zhou, J., et al., 2004. Cloning and characterization of angiocidin, a tumor cellbinding protein for thrombospondin-1. J. Cell. Biochem. 92, 125–146.