MMP-9 gene silencindelivery to maintain

a,1 D. Moya,ald Eras Nhotonicsllergy,, 14203

Nanoparticles have been proposed to act as non-viral gene delivery vectors and have great

of the gene tissue inhibitor ofmetalloproteinase-1 (TIMP-1). This indicates the importance of a

B R A I N R E S E A R C H 1 2 8 2 ( 2 0 0 9 ) 1 4 2 1 5 5

ava i l ab l e a t www.sc i enced i rec t . com

ombalance between the levels of MMP-9 and its natural inhibitor TIMP-1 in maintaining thebasementmembrane integrity. These studies promise the application of a novel nanoparticlebased siRNA delivery system in modulating the MMP-9 activity in BMVECs and other MMP-9producing cells. This will prevent neuroinflammation and maintain the integrity of the BBB.

2009 Elsevier B.V. All rights reserved.potential for therapeutic applications in several disease states. In this study, we evaluated thespecificity and efficiency of quantum dot (QD) complexed with MMP-9-siRNA (nanoplex) indownregulating the expression of MMP-9 gene in brain microvascular endothelial cells(BMVEC) that constitute the BBB. We hypothesize that silencing MMP-9 gene expression inBMVECs and other cells suchas leukocytesmayhelpprevent breakdownof theBBB and inhibitsubsequent invasion of the central nervous system (CNS) by infected and inflammatory cells.Our results show that silencing of MMP-9 gene expression resulted in the up-regulation ofextracellular matrix (ECM) proteins like collagen I, IV, V and a decrease in endothelialpermeability, as reflected by reduction of transendothelial resistance across the BBB in a wellvalidated in-vitro BBB model.MMP-9 gene silencing also resulted in an increase in expression

Keywords:NanotechnologysiRNANanoplexBlood brain barrierMatrix metalloproteinasesMMP-9Quantum dotCNSAvailable online 27 May 20091. Introduction

The blood brain barrier (BBB) plays a cruostasis of the central nervous system (C

Corresponding author.E-mail address: pnprasad@buffalo.edu (P.

1 Equal contribution by both authors.

0006-8993/$ see front matter 2009 Elsevidoi:10.1016/j.brainres.2009.05.047exacerbating neurological diseases such as HIV-1 AIDS dementia and cerebral ischemia.A R T I C L E I N F O

Article history:Accepted 12 May 2009g by a quantum dotsiRNA nanoplexthe integrity of the blood brain barrier

ahajanb,1, Ling Yea, Rajiv Kumara, Hong Dinga,Ravikumar Aalinkeelb, Bindukumar Nairb,. Sykesb, Marco A. Imperialeb, Earl J. Bergeya,. Prasada,

, University at Buffalo, The State University of New York, Buffalo, New York, 14260-3000, USAImmunology, and Rheumatology, University at Buffalo, The State University of New York,, USA

A B S T R A C T

The matrix-degrading metalloproteinases (MMPs), particularly MMP-9, are involved in theneuroinflammation processes leading to disrupting of the blood brain barrier (BBB), therebyAdela Bonoiu , SupriyaKen-Tye Yonga, Indrajit RJessica L. Reynoldsb, DonStanley A. Schwartzb, PaaInstitute for Lasers, Photonics and BiopbDepartment of Medicine, Division of Aand Kaleida Health, Buffalo, New YorkReseach Reportwww.e l sev i e r. ccial role in home-NS), serving as the

N. Prasad).

er B.V. All rights reserved/ l oca te /b ra in resbrain's protective shield against the entry of cells and solublefactors from blood to brain (Ellis et al., 2007). The BBB includesendothelial cells, astrocytes and pericytes, but their interac-tions in the formation or maintenance of the barrier have not

.

modified QDs were fabricated by coating a layer of the

1 2 8been established (Teichberg, 2007). In neuroinflammatoryconditions, such as multiple sclerosis, human immunodefi-ciency virus type-1 (HIV-1) encephalitis (HIVE), meningitis,cerebral ischemia, and brain tumors, the matrix-degradingmetalloproteinases (MMP) play an important role in disruptingthe BBB. MMPs are proteolytic enzymes that are instrumentalin the turnover of the extracellular matrix (ECM), and aremediators of cell migration (Woessner Jr, 1991). Neurons,leukocytes, glial cells, and endothelial cells produce MMPsupon stimulation (Ellis et al., 2007; Paul et al., 1998; Gidday etal., 2005). Furthermore, it has been shown that viral infectionsof the brain enhanceMMP expression several-fold (Sundstromet al., 2001; Wu et al., 2000).

MMP-9 is also known to mediate the transmigration ofinflammatory leukocytes across basement membranes, thusleading to proteolytic degradation of extracellular matrix(ECM) components. In addition to degrading the neurovascu-lar matrix, MMP-9 promotes neuronal apoptosis by disruptingcellmatrix interactions that enhance the permeability of theBBB, thereby exacerbating neurological diseases such as HIVE.Studies with MMP-9 knockout mice have shown that dimin-ished MMP-9 expression reduces BBB leakage (Asahi et al.,2001). In vitro studies using brain endothelial cell culturesactivated with the pro-inflammatory cytokines, tumor necro-sis factor-alpha (TNF-) and interleukin-1beta (IL-1), resultedin selective up-regulation of MMP-9 activity, whereas nosignificant changes were seen in the MMP-2 or TIMP-2 levels(Persidsky and Gendelman, 1997). This suggests that MMP-9 isa predominant MMP in neuroinflammatory processes leadingto BBB leakage (Persidsky et al., 1997).

To invade the CNS, HIV-1 must migrate through the brainmicrovascular endothelial cells (BMVECs) that are a componentof the BBB (Pulliam et al., 1997; Schmidtmayerova et al., 1997).ThepathogenesisofHIVE isprimarilyattributed to infiltrationofthe CNS by HIV-infectedmononuclear cells. HIVE often exhibitsimpairment of the BBB integrity (Persidsky and Gendelman,1997, 2003; Annunziata, 2003; Langford et al., 2006).

The inhibition of MMP-9 activity via post-translationalgene silencing using delivery of short interfering RNA (siRNA)molecules at BMVECs will have a significant impact onreducing the BBB permeability. However, a major hurdlelimiting the use of this technology is the lack of methods tosafely and efficiently deliver siRNA molecules to their target(Whitehead et al., 2009). In the free form, siRNA moleculeshave a very short half life in physiological conditions, owing totheir vulnerability for degradation by endogenous nucleases.Therefore, they need nanoparticulate carriers (vectors) thatwill not only protect them from degradation in the biologicalmilieu, but also steer them to desired cells/tissues as well asfacilitate their cellular entry (Whitehead et al., 2009). The useof semiconductor nanocrystals (NCs) or so-called quantumdots (QDs) as luminescence probes for numerous biologicaland biomedical applications has become an area of intenseresearch over the last decade (Michalet et al., 2005; Prasad,2003; Prasad, 2004).

QDs offer several advantages over organic dyes, includingincreased brightness, stability against photobleaching, broadabsorption spectra, and a tunable and narrow emission

B R A I N R E S E A R C Hspectrum (Prasad, 2004). As a result, they have demonstratedthe potential to dramatically outperform conventional organicamine functionalized polymer polydiallyldimethylammo-nium chloride (PDDAC) on the surface of carboxyl-terminatedQDs. Two different QD:PDDAC ratios were used, representedas QD-15P and QD-30P, with the later containing twice theamount of the polymer than the former for the same amountof QDs (details provided in Experimental procedures). Thephysical and spectral characterization of the electrostaticcomplexes of QDs with siRNA was first determined usingfluorescently labeled siRNA (siRNAFAM). The fluorescentlylabeledmarker siRNAwas designed for monitoring the uptakeof siRNA by fluorescence microscopy or other fluorescencebased technique, since it contained the fluorescein derivativeFAM (5-carboxyfluorescein) on the 5 end.

To examine the binding efficiency of nanoplexes, weperformed agarose gel electrophoresis using siRNAFAM (greenemitting) complexed with the two formulations of QDs (QD-15P and QD-30P, both red emitting). As shown in Fig. 1A, freesiRNAFAM appeared as a leading band in the gel (lane 1). For theQD-only (QD-15P and QD-30P) bands, corresponding to lanes 2and 4, respectively, all the fluorescence was observed in thewells, indicating that the QDs did not migrate into the gel.However, upon comparing the bands from the two nano-plexes, we found that while the green band for QD-15PsiRNAFAM is slightly diminished (lane 3), it has completelydyes in imaging of cellular and sub-cellular structures and in avariety of bioassays (Prasad, 2003; Prasad, 2004; Qian et al.,2007; Yong et al., 2007). The surface of the QDs can be easilyfunctionalized to incorporate biologicals such as siRNAs,aptamers, antibodies, peptides, and proteins (Michalet et al.,2005). Furthermore, the size of these nano-bioconjugatesunder physiological condition is typically less than 50 nm,making these QD-based nanoparticles ideal candidates for invitro and in vivo studies and applications.

In this investigation, we evaluate the specificity andefficiency of QDs as non-viral gene nanocarriers for MMP-9gene silencing in a primary brain endothelial cell line that iswell known to be difficult to transfect. Additionally, wemeasured functional/physiological changes as a consequenceofMMP-9 gene silencing in these BMVEC.We hypothesize thatsilencing of the MMP-9 gene expression in BMVEC willmodulate the expression of ECM proteins like collagens I, III,IV, and V, and decrease endothelial permeability as reflectedby reduction of transendothelial electrical resistance acrossthe BMVEC monolayer. These studies will not only highlightthe role of MMP-9 in the modulation of BBB permeability, butalso validate a novel nanoparticle based system for efficientsiRNA delivery.

2. Results

2.1. Complex formation of siRNA with theQDs (nanoplexes)

Initially we determined the optimal conditions for theformation of QDsiRNA nanoplexes. For electrostatic com-plexation with the anionic siRNA molecules, cationically

1432 ( 2 0 0 9 ) 1 4 2 1 5 5disappeared in the case of QD-30PsiRNAFAM (lane 5). Thisimplies that the siRNA was partially and completely com-

(siRNAMMP-9). Since siRNAMMP-9 was non-fluorescent, this gelwas stained with Ethidium Bromide (EtBr) for visualizing thesiRNA (Fig. 1B). The same outcome was observed as in Fig. 1A,with the siRNA molecules partially (lane 3) and completely(lane 5) complexed by QD-15P and QD-30P, respectively. Datapresented in Fig. 1 is a representative gel from 3 separateexperiments.

2.2. Size and spectral characterization

Transmission electron microscopy (TEM) images of QD-30PsiRNAFAM nanoplexes showed that they remain as individualparticles with a core diameter of 1520 nm (Fig. 2A). It can bevisualized from the TEM that siRNA complexation did notcause any aggregation of the QDs and this colloidal stabilitywas observed for more than a month after complexation.Upon recording the emission spectra of the QD-30PsiRNAFAM

FAM

144 B R A I N R E S E A R C H 1 2 8 2 ( 2 0 0 9 ) 1 4 2 1 5 5Fig. 1 Effective binding of siRNA to QD as determined byagarose gel electrophoresis. A. Gel image with no ethidiumbromide staining. lane 1: 50 pmol free siRNAFAM; lane 2: 15Pplexed with the QDs in the cases of QD-15PsiRNAFAM and QD-30PsiRNAFAM, respectively. These results suggest that the QD-30P was the more efficient formulation of the two for siRNAcomplexation. It is worth noting though that the fluorescencefrom the siRNAFAM was not apparent at the origin (well) of thegel owing to the overwhelming fluorescence signal emitted bythe QDs. We have repeated this study with the sameexperimental set-up, except for using siRNA against MMP-9

Fig. 2 QDsiRNAFAM nanoplex characterization: A. Transmissioaverage diameter of the nanoplex is 15 nm (bar scale=30 nm). NB. Emission spectra of free siRNAFAM, free QDs, and QDsiRNAFA

QD; lane 3: QD-15PsiRNAFAM nanoplex; lane 4: 30P QD; lane5: QD-30PsiRNAFAM nanoplex. B. Ethidium bromide stainedgel image. All lanes are identical to A, except that the siRNAused was not FAM labeled and was specific for MMP-9.nanoplexes, as well as that of the 30P-QDs and siRNAindividually, it was observed that the red emission (max620 nm) from the QDs increases at the cost of the greenemission (max 535 nm) from the siRNAFAM upon complexa-tion (Fig. 2B). This was attributed to the process of fluores-cence resonance energy transfer (FRET) from the siRNAFAM

(donor) to the QDs (acceptor), owing to the overlappingabsorption spectra of the later with the emission peak of theformer. This process of FRET is possible only when the twointeractive species are in close physical proximity, in the rangeof a few nanometers from each other. Therefore, the observa-tion of FRET in this case further substantiates complexformation between these two oppositely charged species.

2.3. Cellular uptake of fluorescent siRNA afterQD-mediated delivery

Following the confirmation of the complexation of thesiRNAFAM with the QDs, the cellular uptake and intracellulardistribution of the QD-15PsiRNAFAM and QD-30PsiRNAFAM

nanoplexes were studied using confocal microscopy. Fig. 2shows emission from brain microvascular endothelial cells(BMVECs) treated with the nanoplexes, with the green

n electron microscope image of nanoplexes showing that the

o aggregation was observed after nanoplex formation.M nanoplexes.

emission coming from siRNAFAM, red emission coming from theQDs,and theyellowemission isa resultofanoverlapof these twofluorescent species. However, in contrast, cells treated with freesiRNAFAM showed no emission at all. This demonstrates that theintracellular delivery of siRNA was mediated by the presence ofQDs as carriers. Also, the superior siRNA complexation (Fig. 1)and intracellular uptake (Fig. 3) of the QD-30PsiRNAFAM nano-plexes over that of the QD-15PsiRNAFAM nanoplex suggests thatthe former nanoplex formulationwas better suited over the laterformulation for siRNAdelivery. Therefore, in subsequent studies,only the QD-30PsiRNAFAM or QD-30PsiRNAMMP-9 formulationwere used, which are hereafter referred to as QDsiRNA. Datapresented in Figs. 2 and 3 are representative pictures from 3separate experiments.

2.4. Effect of nanoplexes on cell viability

We evaluated the effect of QD complexed with siRNA, specificfor thematrixmetalloproteinase-9 (MMP-9) gene (siRNAMMP-9),on the viability of BMVEC using the MTT assay (Promega,Madison, WI). We observed that BMVEC treated with QD

siRNAMMP-9 nanoplexes remained >95% viable under alltreatment conditions for up to 2 weeks (data not shown).This result demonstrates that the QD-mediated delivery ofsiRNA did not produce any appreciable cytotoxicity.

2.5. Efficiency of QDsiRNAMMP-9 nanoplexes onknockdown of MMP-9 gene expression by BMVEC

After confirming the efficient QD-mediated cellular delivery ofthe marker siRNAFAM (Fig. 2) and its non-cytotoxicity, weinvestigated functional gene silencing efficiency using nano-plexes with siRNAMMP-9. The efficiency of gene silencing wasdetermined by measuring the percentage inhibition of theexpression of the MMP-9 gene using quantitative, real-timePCR (Q-PCR). A fixed concentration of siRNAMMP-9 (200 pmol)was complexed to QDs. Our results (Fig. 4) show the decreasein MMP-9 expression by BMVEC treated with QDsiRNAMMP-9

nanoplexes to be 78% [transcript accumulation index(TAI)=0.220.03, p=0.0001], when compared to the negativecontrol (Xtreme transfection reagent plus scrambled siRNA[200 pmol]) (TAI=1.000.03). The positive control (Xtreme

d Q

145B R A I N R E S E A R C H 1 2 8 2 ( 2 0 0 9 ) 1 4 2 1 5 5Fig. 3 Confocal images of BMVEC treated with siRNAFAM an

QD-15PsiRNAFAM, (B) QD-30PsiRNAFAM, and (C) free siRNAFAM. Poverlays of 2 fluorescence (QD and siRNAFAM) and transmission iDsiRNAFAM nanoplexes: BMVECs treated with (A)

anel L: fluorescence (green) due to siRNAFAM. Panel M:mages, and Panel R: fluorescence (red) due to QDs only.

produced a greater suppression of the MMP-9 activity than thatobtained using the commercially available transfection agent,Xtreme.Thus,QD-siRNAcanbeconsidered tobehighly effectiveagent for gene silencing.

2.7. Effect of MMP-9 gene silencing on the expression ofTIMP-1 and EMMPRIN genes

MMPs are selectively inhibited by tissue inhibitors of metallo-proteinases (TIMPs). TIMP-1 binds to both the active andprecursor form of MMP-9, and inhibits its enzymatic activity.In neuroinflammatory conditions, there is a reciprocal regula-tion of the expression of levels of MMP-9, which is elevated,and TIMP-1, which is decreased, in BMVECs. MMP activationalso results in increased levels of the extracellular matrixmetalloproteinase inducer (EMMPRIN). EMMPRIN is a 57 kDtransmembrane glycoprotein that can stimulate production ofvarious MMPs (Foda and Zucker, 2001). EMMPRIN is believed tohave a role in physiological tissue remodeling by inducingstromal MMPs. However, in the BBB, EMMPRIN could conse-quently result indamage to thebasal lamina, therebydisruptingthe BBB. When BMVEC were treated with QDsiRNAMMP-9

nanoplexes, we observed a significant increase in the TIMP-1gene expression (p

synthesis. Our results show a decrease in the MMP-9 proteinexpression in BMVECs treated with QDsiRNAMMP-9 (Figs. 7Aand B). Further, an increase in the TIMP-1 expression in theQDsiRNAMMP-9 transfected cells was observed (Figs. 7C andD).

In addition to immunofluorescent staining, we quantified theprotein levelsofMMP-9andTIMP-1usinganELISAkit.Cell culturesupernatants from 4 separate experiments, where BMVEC cellswere transfected with the transfection agent Xtreme complexedwith siRNA against MMP-9 (XisiRNAMMP-9) and QD conjugatedsiRNAnanoplexes, (15P)QDsiRNAMMP-9 and (30P)QDsiRNAMMP-9

respectively were harvested and the MMP-9 and TIMP-1 proteinlevels were quantified using commercially available ELISA kits (R& D Systems; Minneapolis, MN). Our results (Fig. 8) show thatMMP-9 protein levels (ng/ml) for cells treatedwithXisiRNAMMP-9,(15P) QDsiRNAMMP-9 and (30P) QDsiRNAMMP-9 were 10.011.26(p=0.007), 6.401.56 (p=0.004), and 4.241.36 (p=0.005), respec-tively and were significantly lower than the untreated control(14.421.99 ng/ml). On the other hand, the TIMP-1 protein levels(ng/ml) for cells treatedwith XisiRNAMMP-9, (15P) QDsiRNAMMP-9

and (30P) QDsiRNAMMP-9 were 3.101.61 (p=0.03), 5.841.90(p=0.007), and 8.071.88 (p=0.003), respectively and were sig-nificantly higher than the untransfected control (0.350.25 ng/ml). The significant modulation of the expression of MMP-9 andTIMP-1 protein as a consequence of MMP-9 gene silencing inBMVEC demonstrates that there is a dynamic balance between

Fig. 7 Immunofluorescent staining forMMP-9 andTIMP-1 in unexpression in untransfected BMVEC. B. MMP-9 expression in QDexpression in untransfected BMVEC; D. TIMP-1 expression in QD

Fig. 6 Effect of QDsiRNAMMP-9 nanoplexes on theexpression of TIMP-1 and EMMPRIN genes in BMVEC. BMVECwere treated in vitro with either QDsiRNAMMP-9 nanoplexes,the positive transfection control (XtremesiRNAMMP-9), or thenegative transfection control (XtremesiRNAscrambled) for 48 h.RNA was extracted, reverse transcribed, and the resultantcDNAwas assayed for TIMP-1 and EMMPRINgene expressionby real-time, quantitative PCR. Relative expression of mRNAspecies was calculated using the comparative CT method.Data are the meanSD of 3 separate experiments done induplicate. Statistical significance was determined usingANOVA based on comparison between the QDsiRNAMMP-9

nanoplexes, the positive transfection control and thenegative control respectively.

147B R A I N R E S E A R C H 1 2 8 2 ( 2 0 0 9 ) 1 4 2 1 5 5the expressions of these inter-related genes and proteins.

2.9. Effect of MMP-9 silencing on the gene and proteinexpression of Collagens I1, III, IV and V by BMVEC

Collagens are the major component of the basal laminawhich is essential for maintenance of the endothelial

transfected andQDsiRNAMMP-9 transfected BMVEC. A.MMP-9

siRNAMMP-9 nanoplex transfected BMVEC. C. TIMP-1siRNAMMP-9 nanoplex transfected BMVEC.

was increased 77% (p

1 2 8B R A I N R E S E A R C Htransfected, scrambled siRNA negative controls (20214.14 /cm2). These data suggest that silencing of the MMP-9 geneexpression in BMVEC may be useful for the treatment of avariety of neuroinflammatory states that may compromisethe integrity of the BBB.

Fig. 10 Effect of the QDsiRNAMMP-9 nanoplex on Collagen 11,analysis of the lysates from transfected and untransfected BMVEsiRNAMMP-9 nanoplexes (Lane 4); (30P) QDsiRNAMMP-9 nanoplexes(Lane 3) and negative control (XtremesiRNAscrambled) (Lane 2), atuntransfected control (Lane 1). Panel A shows data from a represb-actin (42 kDa band) protein expression, and (ii) a significant inccells treated with the 15P and 30P QDsiRNAMMP-9 nanoplexes (Lacontrol (Lane 1). (iii) a significant increase in Collagen III (2115 k30P QDsiRNAMMP-9 nanoplexes (Lane 4 and 5), as opposed to celldecrease in Collagen IV (2170 kDa band) protein expression, in c(Lane 4 and 5), as opposed to cells treatedwith untransfected contband) protein expression, in cells treated with the 15P and 30P Qtreatedwith untransfected control (Lane 1). Panel B: Graphical repbands showing percentage change in the protein expression for arespectively in transfected cells as compared to the untransfectedexperiments, and statistical significance was determined using A1492 ( 2 0 0 9 ) 1 4 2 1 5 53. Discussion

The pathogenesis of HIVE (Persidsky and Gendelman, 1997,2003) is associated in part with BBB breakdown and infiltration

III, IV and V protein expression in BMVEC cells. Western blotC cells following their treatment in vitro with (15P)QD(Lane 5), aswell as both positive control (XtremesiRNAMMP-9)92 h post-transfection and comparisons made with anentative western blot experiment showing (i) no change inrease in Collagen 11 (290 kDa band) protein expression, inne 4 and 5), as opposed to cells treated with untransfectedDa band) protein expression, in cells treated with the 15P ands treated with untransfected control (Lane 1). (iv) a significantells treated with the 15P and 30P QDsiRNAMMP-9 nanoplexesrol (Lane 1). (v) a significant increase in Collagen V (2150 kDaDsiRNAMMP-9 nanoplexes (Lane 4 and 5), as opposed to cellsresentation of the densitometric analysis of the all the protein) Collagen 1a1; b) Collagen III; c) Collagen IV and d) Collagen Vcontrols. Data shown aremeanSD of results from 3 separateNOVA. * indicates p

1 2 8of HIV-infected mononuclear cells into the CNS. Increasedpermeability of the BBB is an early and critical event that often

Fig. 11 MMP-9 gene silencing by QDsiRNAMMP-9

nanoplexes affects the functional integrity of an in-vitro BBBmodel. BMVEC, untransfected or transfected withQDsiRNAMMP-9 nanoplexes, were used to form a model BBBin vitro. The transendothelial electrical resistance (TEER) wasmeasured 48 h post-transfection. Using QDsiRNAMMP-9

transfected BMVEC to form the BBB resulted in significantlyhigher TEER values compared to untransfected orQDsiRNAscrambled negative controls. Data shown are themeanSD of 3 separate experiments. Statistical significancewas determined using ANOVA based on comparisonsbetween transfected cells and controls.

150 B R A I N R E S E A R C Hprecedes symptoms of HIVE. MMPs are a family of zymogenproteases, which under normal physiological conditions areinvolved in the degradation and remodeling of the extra-cellular matrix. MMPs degrade the basal lamina proteins,fibronectin, laminin and collagens, and also modulate thetight junction proteins (TJPs). Subsequent proteolysis of themicrovascular basal lamina thus causes a breakdown of theBBB (Abbott, 2005; Abbott, 2004; Paul et al., 1998). Althoughneurons produce MMP-9, within the CNS, leukocytes are themajor cellular source of MMP-9, which in turn facilitatesfurther leukocyte recruitment (Gidday et al., 2005). Thepassage of leukocytes through basement membranes involvesproteolytic degradation of extracellular matrix (ECM) compo-nents. Matrix metalloproteinase (MMP)-9 mediates transmi-gration of inflammatory leukocytes across basementmembranes and this has significant relevance in severalpathological conditions such as cancer, allergic asthma andneurological disorders. Localized tissueMMP-9 levels also playan important role in cell trafficking. MMP-9 gene deletion inendothelial cells will cause a reduction in BBB disruption bymodulation of tight junction proteins and also reduceleukocyte trafficking across the BBB.

A new therapeutic intervention, silencing MMP-9 geneexpression in cells that constitute the BBB, may be useful inneuroinflammatory conditions such as HIVE, to preventbreakdown of the BBB and inhibit subsequent invasion ofthe CNS by HIV-infected leukocytes. Several strategies fordelivery of siRNA using nanoparticles have been investigated,including encapsulation/complexation of siRNA in lipid andpolymer based nanomaterials. However, the larger size(diameter above 200 nm) and potential antigenicity of suchmaterials have limited the progress of siRNA technology intothe clinic. We have synthesized ultrasmall QD bioconjugates(diameter 1520 nm, Fig. 2) which are water-dispersible,photostable, and easily form stable electrostatic complexeswith siRNA molecules. The efficient complexation of siRNAwith the QDs and their cellular delivery was confirmed usingfluorescently labeled siRNA (Figs. 1 and 2). Additionally, ournanoplexes manifested no toxic effects on cells and >95%viability was observed (data not shown).

Using siRNA against MMP-9, the QD-mediated genesilencing efficiency of almost 80% was obtained in BMVECs(Fig. 4). Zymography, immunofluorescence studies andprotein quantification using ELISA methodologies alsocorrelate this significant decrease in MMP-9 expression inBMVECs mediated by QDsiRNAMMP-9 (Figs. 58, respec-tively). We observed that MMP-9 gene silencing in BMVECby QDsiRNAMMP-9 also affected the expression of other genes.TIMP-1 expressionwas significantly up-regulated in BMVEC byQDsiRNAMMP-9 (Figs. 7 and 8). Under physiological conditionsthe activity of MMP-9 is regulated on several levels, includinggene transcription, activation of inactive precursors, andmodulation of TIMPs. A down regulation in MMP-9 activitywas observed in BMVEC in whichMMP-9 gene expression wassilenced using the nanoplexes. However a similar magnitudeof decrease in MMP-2 activity was not observed in these cells,whichmay be attributed to the fact that despite the functionalsimilarities of MMP-2 and MMP-9, the promoters for MMP-2and -9 are structurally distinct (Bergman et al., 2003). Theactivation of MMPs can occur within the cell, at the cell surfaceor extracellularly by oxidative stress (Wang et al., 2002), andunder conditions of oxidative stress the full length form of theMMP-2 may also be proteolytically active. In the currentexperimental paradigm, cells were not undergoing anyoxidative stress, and therefore we did not see intenseproteolytic activation of MMP-2.

MMP-9 is secreted as pro-MMP-9, and requires activationby proteolytic enzymes for its catalytic function. The molarratio of endogenous MMP-9 to TIMP-1 dictates whether pro-MMP-9 activation can progress. When the MMP-9/TIMP-1ratio exceeds 1.0, MMP-9 activation occurs through aprotease cascade involving plasmin and stromelysin 1(MMP-3) (Ramos-DeSimone et al., 1999). Thus there is adynamic balance between the expression of MMPs andTIMPs. Any perturbation of this equilibrium, such as silen-cing of MMP-9 expression may secondarily affect TIMP-1expression, as we observed (Figs. 68) particularly since it isbelieved that the activity of the MMPs may be modulated bythe endogenous TIMPs.

Collagens are the principal substrate of MMPs. The sumbalance of collagen deposition is a function of both theirsynthesis and degradation. We observed that silencingMMP-9gene expression in BMVEC also caused reciprocal increases inthe expression of various collagens, notably I1, IV and V (Figs.9 and 10). MMPs are best known for their actions in ECMremodelling. MMPs degrade collagens I, III, IV and V which aretheir natural substrates. All of these genes are inter-relatedand are involved in remodeling of the ECM. We noted a

2 ( 2 0 0 9 ) 1 4 2 1 5 5significant increase in the gene and protein expression ofcollagens I1 and V, and although we did not observe a

1 2 8significant increase in Collagen III gene expression, a sig-nificant increase in the protein levels of Collagen III wasobserved in theMMP-9 transfected cells, further both gene andprotein expression levels of Collagen IV were significantlydecreased in MMP-9 transfected cells. Collagen I1 andCollagen III are both not substrates for MMP-9, while CollagenIV is a MMP-9 substrate. Collagen type IV represents 50% ofall basementmembrane proteins, and has the capacity to self-assemble into organized networks. Unlike fibrillar collagens oftypes I, II, III, and V, type IV collagen is found to be crucial forbasement membrane stability and assembly (Timpl, R et al.,1981). Since, Collagen type IV is a keyMMP-9 substrate, it is notsurprising that a suppression of MMP-9 gene expressionresulted in a significant decrease in Collagen type IV expres-sion. The other collagen types I, III, and V that we quantified inthe study, were increased in BMVEC in which MMP-9 expres-sion was inhibited using the siRNA nanoplexes. This may beattributed to the fact that inhibition of MMP activity isassociated with less deposition of fibrotic tissue, while thetotal matrix collagen content is a function of both synthesisand degradation, the degraded products of matrix proteinsmay serve as stimulators for collagen synthesis (Maquartet al., 1988; Li et al., 2000). The changes in gene and proteinexpression of these family of Collagens observed, may besecondary to changes in a well orchestrated, complex,dynamic, network of inter-related genes and proteins involvedin the recurring changes in the ECM that occur in both healthyand diseased states.

EMMPRIN is involved in the breakdown of the BBB byinducing nuclear factor (NF)-B that, in turn, increases theexpression of the proinflammatory cytokines, IL-6 and TNF-(Schmidt et al., 2008). NF-B binding by inflammatorycytokines is necessary for the up-regulation of MMP-9. TheMMP-9 promoter has an upstream nuclear NF-B site that iscrucial to its activity (Bond et al., 1998). However, no changesin the gene expression of EMMPRIN were observed in BMVECtreated with QDsiRNAMMP-9. This is not a surprising finding,as no stimuli for inflammation were used in our system,thus no effects on EMMPRIN expression were anticipated.Moreover, the lack of an effect on EMMPRIN expressionserves as a good, negative control for the effects of QDsiRNAMMP-9 on the gene expression of TIMP-1 and thecollagens. Thus, the up-regulation of TIMP-1 and collagengene expression is highly selective. This suggests that thereis a balance in the expression of various genes involved intissue remodeling, and any changes in the expression onealters the homeostasis between them, resulting in concomi-tant changes in the expression of other functionally relatedgenes. Our gene, protein expression and immunofluores-cence studies also show a significant increase in the TIMP-1expression in BMVEC transfected with QDsiRNAMMP-9.Levels of TIMP-1 in BMVEC are decreased under neuroin-flammatory conditions in which the MMP-9 levels areincreased. An increase in the TIMP-1 expression in BMVECtransfected with QDsiRNAMMP-9 is likely due to a responseto an alteration in the normal MMP/TIMP ratio. This confirmsthe importance of the balance between the levels of MMP-9and its natural inhibitor TIMP-1 in maintaining the basement

B R A I N R E S E A R C Hmembrane integrity. Although, the TIMPs in general do notdemonstrate specificity for any particular MMP, TIMP-2 hasbeen shown to have some degree of preference for MMP-2and TIMP-1 for MMP-9 (Brew et al., 2000; Goldberg et al.,1992). Thus, MMP-9 modulation resulted in a significantincrease in TIMP-1 gene and protein expression, and whilewe did not measure the levels of TIMP-2, we can speculatethat the levels of TIMP-2 may not have changed significantlyin these cells, thereby not altering the MMP-2 activitysignificantly as indicated in our zymography studies.Although it is difficult to speculate on the precise mechan-isms responsible for increased TIMP-1 gene and proteinexpression in the BMVEC cells in which MMP-9 geneexpression was silenced, we believe that it was the resultof a consequent perturbation in the balance between theexpression of MMP-9 and TIMP-1.

We did not observe any significant change in EMMPRINexpression in BMVEC transfected with QR-siRNAMMP-9. EMM-PRIN stimulates MMP-9 in monocytes (Burggraf et al., 2005;Caudroy et al., 1999). Thus, the effect of MMP-9 gene knock-down on the expression of other genes is not indiscriminateand is highly specific. We are currently examining how thesecomplicated inter-related pathways are regulated. The abovefindings are consistent with the premise that many genesinvolved in ECM remodeling also regulate the expression ofeach other. Further, based on the results of the TEERmeasurements across the in vitro BBB (Fig. 11) we canspeculate that MMP-9 is a key factor in the multiple mechan-isms that may be involved in the breakdown of the BBB duringneuroinflammatory processes and thus can be a major targetfor subsequent nanotherapy.

Nanoplexes provide an advantage over current therapeuticoptions, with respect to drug targeting, delivery and release,due to their ability to combine diagnosis and therapy. Thegoals of nanotherapy include improving the stability ofnanoplexes in the biological environment; the capacity tomediate the bio-distribution of active compounds; and theability to load, transport, target, and release their therapeuticpayloads. Although many previously described nanoparticleshave been found to be cytotoxic, further research has led tobetter tolerated formulations as we describe herein. Subse-quent clinical trials are necessary to assure that our nanoplexconsisting of QDs and siRNAMMP-9 is both safe and effective forthe therapy of neuroinflammatory disorders that disrupt theintegrity of the BBB.

Albeit this manuscript describes a novel QR conjugatedMMP-9 siRNA nanoplex that specifically targets MMP-9gene expression in BMVECs that form the blood brainbarrier, the present technology could be used to targetthe MMP-9 gene in various cell types such as leukocytes.This will help in modulating MMP-9 production under avariety of patho-physiological conditions. The currentstudy, therefore, describes a general method for theknockdown of MMP-9 gene expression that could poten-tially have therapeutic application beyond the currentarea of investigation.

In conclusion this study supports the premise that MMP-9plays a key role in BBB disruption during neuroinflammation,and that use of novel nanotechnology delivery system tosilenceMMP-9 expression in cells that constitute the BBB, may

1512 ( 2 0 0 9 ) 1 4 2 1 5 5help preserve the integrity of the BBB and may be useful inclinical settings associated with BBB disruption.

1 2 84. Experimental procedures

4.1. Nanoparticle formulation

Cadmium oxide (CdO), oleic acid, selenium (Se), trioctylpho-sphine (TOP), diethylzinc (Et2Zn), hexamethyldisilathiane[(TMS)2S], chloroform, ammonium hydroxide, mono-N-(t-Boc)-propylenediamine, trifluoroacetic acid, lysine, mercapto-succinic acid (MSA), mercaptoundecanoic acid (MUA) werepurchased from Aldrich. All chemicals were freshly preparedfor each new set of experiments. Semiconductor nanocrystals,quantumdots (CdSe/CdS/ZnS), are prepared by growing a CdS/ZnS graded shell on CdSe core in non-aqueous medium.Details of the preparation procedure are described in ourprevious publication (Qian et al., 2007). MSA was used totransfer the QDs from the non-aqueous phase to the aqueousphase, leading to the formation of water-dispersible carboxyl-terminated QDs. In order to fabricate a net positive charge onthe QD surface, the surface of the carboxyl-terminated QDswere further coated with a layer of the cationic, aminefunctionalized polymer poly(diallyldimethylammoniumchloride (PDDAC) in two different QD:PDDAC ratios (repre-sented as 15P and 30P, with the PDDAC amount twice in thelater as compared to the former, for a fixed QD amount). Thewater-dispersible, cationically modified QDs were then elec-tostatically attached to the anionic siRNA molecules. Theresultant nanoplexes (QDsiRNA) are prepared in freshOptiMEM media (Invitrogen Corporation, Carlsbad, CA) beforethe start of each experiment.

4.2. Physical characterization of QD

The particle size of our nanoparticle formulations is deter-mined by transmission electron microscopy (TEM). An aqu-eous dispersion of the nanoparticles is dried on formvar-coated copper grids obtained from Electron MicroscopySciences Inc. and visualized using a JEOL JEM 2020 electronmicroscope. Simple spectrophotometry and spectrofluorime-try are used to study the spectral properties of the fluorescentnanomaterials. UVVis absorption spectra are recorded with aShimadzu UV-3101 PC spectrophotometer using a quartzcuvette with a 1 cm path length. Fluorescence spectra arerecorded on a Fluorolog-3 spectrofluorimeter (Jobin Yvon,Longjumeau, France).

4.3. Cell culture

Primary cultures of human brainmicrovascular endothelial cells(BMVEC) and normal human astrocytes (NHA) are obtained fromApplied Cell Biology Research Institute (ACBRI) Kirkland, WA.Characterization of these cells demonstrated that >95% ofBMVECs were positive for cytoplasmic VWF/Factor VIII and>99% of the NHA were positive for glial fibrillary acidic protein(GFAP). BMVEC are cultured in CS-C complete medium (ABCRI)with attachment factors (ABCRI) and Passage Reagent Group(ABCRI). NHA are cultured in CS-C complete serum-freemedium(ABCRI), supplemented with 10 g/ml human epidermal growthfactor (EGF), 10 mg/ml insulin, 25 g/ml progesterone, 50 mg/ml

152 B R A I N R E S E A R C Htransferrin, 50 mg/ml gentamicin, 50 g/ml amphotericin, and10% fetal bovine serum (FBS). Both BMVEC andNHA are obtainedat passage 2 for each experiment and are used for all experi-mental paradigms between 2 and 8 passages, within 6 to 27cumulative population doublings.

4.4. Cell viability assay

This assay measures the reduction of a tetrazolium component3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide(MTT) into an insoluble formazan product by the mitochondriaof viable cells. The MTT assay is a quantitative and sensitivemeasure of cell proliferation, since it determines the growth rateof cells relative to a linear relationship between cell number andabsorbance. Cells (1104/ml) transfected with nanoplexes areincubatedwith theMTTreagent for3h, followedby additionofa detergent solution to lyse the cells and solubilize the coloredcrystals. The samples are read in an ELISA plate reader at awavelength of 570 nm. All the samples used are prepared intriplicate and values mean standard deviation are calculated.

4.5. Determination of the delivery of nanoplexesinto BMVEC

Nanoplexes (QDsiRNA) are discrete spheres, having a meansize of 1520 nm. They are sufficiently small to diffuse into cellsby endocytosis and favor nucleic acid transfer and release fromtheendosomeacross thenuclearmembrane. Nanoplex suspen-sions are prepared by adding 100 pmol siRNAFAM (fluorescentsilencer-Ambion) and 20 L of QD nanoparticles (15P and 30Pformulations) to 200 L media without serum; they areincubated for 10 min at room temperature and added to35 mm culture plates. Commercial transfection reagent(Xtreme, Roche Applied Biosystem) is used as a control totransfect 100 pmol siRNAFAM into BMVEC according to themanufacturer's protocol. After 24 h post-transfection, themedium is replaced by fresh OptiMEM containing 10% FBS,and cells are cultured in an incubator (VWR Scientific, model2400) at 37 C in a humidified atmosphere of 5% CO2 in air. Forconfirming the consistency of the results, samples are preparedin triplicate. Fluorescence images are acquired after 24 h with aspectral confocal microscope (TCSSP2, Leica MicrosystemsHeidelberg GmbH) with a HXC PL APO CS 63.01.40 oilimmersion objective. The samples are excited by a laser at488 nm (PDL800-D, PicoQuant GmbH). For siRNAFAM detection,the photomultiplier tube (PMT) 1 is set with a filter of 520540 nm for the detection of FAM, and PMT2 is set with a filter of620680nmfor thedetectionofQDbioconjugates. ForQDsiRNAnanoplexes, PMT1 is set with a filter of 450550 nm forautofluorescence, the PMT2 is set with a filter of 580620 nmfor QD (608 nm), and PMT3 is set with a filter of 640690 nmQDs(608 nm). A water-immersion objective lens (Nikon, Fluor 60,NA1.0) is used for the cells.

4.6. Real-time, quantitative PCR (Q-PCR)

Total RNA is extracted from BMVEC using Trizol reagentaccording manufacturer's protocol, quantified by spectropho-tometry (Nano-Drop Wilmington, DE) and stored at 80 Cuntil used. Q-PCR is used to quantitate MMP-9 and collagen IV

2 ( 2 0 0 9 ) 1 4 2 1 5 5expression inBMVEC cultures. Approximately 1106 BMVECareharvested at 48 h post-transfection, and RNA is extracted as

1 2 8described above. The RNA is then reverse transcribed to cDNAusing a reverse transcriptase kit (Promega Inc, Madison, WI).Relative abundance of each mRNA species is quantitated by Q-PCR using specific primers and the Brilliant SYBR green Q-PCR master mix (Stratagene Inc, La Jolla, CA). To provideprecise quantification of the initial target in each PCR reaction,the amplification plot is examined and the data are interpretedas described (Bustin, 2002; Radoni et al., 2004). Data are themeanSD of 3 separate experiments done in duplicate.Statistical significance was determined using ANOVA basedon comparisons between QDsiRNAMMP-9 nanoplexes and thenegative control.

4.7. Western blot analysis

Protein was extracted from transfected cells and untrans-fected controls using the M-PER protein extraction reagent(Cat # 78503, Pierce, Inc, Rockford, IL). Immunodetection usingstandard Western blotting techniques as described in theCurrent Protocols in Immunology were used. 50 g totalprotein extracted from cells was loaded per lane andseparated by a 420% SDS-Tris glycine PAGE. Gels are blottedonto nitrocellulose membranes and membranes are probedwith the Collagen I1, III, IV and V (Cat #'s sc-8788, sc-8781, sc-9302 and sc-9581 respectively from Santacruz Biotech; Santa-cruz CA) antibodies.

4.8. Measurement of MMP-9 and TIMP-1 proteinexpression using ELISA

Cell supernatants from transfected cells and untransfectedcontrols were harvested 96 h post-transfection and the levels ofMMP-9 and TIMP-1 were measured in the supernatants usingcommercially available Quantikine-colorimetric sandwichELISA kits [MMP-9 assay (Cat # DMP900) and TIMP-1 assay (Cat# DTM100)] from R andD systems, Minneapolis, MN. Briefly, thekit constitutes monoclonal antibody pre-coated microplatesthat are specific for MMP-9 or TIMP-1. Standards and samplesare pipetted into the wells, followed by an incubation intervaland then washing. An enzyme linked polyclonal antibodyspecific to the particular ligand is then added, followed by awashing step that ensures removal of unbound antibodyenzyme reagent. A colorimetric detection is done using thestabilized chromogen TMB (tetramethybenzidine). The mini-mum detectable dose (MDD) dose of the MMP-9 assay was0.156 ng/ml, and the standard curve ranged from 0 to 20 ng/ml.The minimum detectable dose (MDD) dose of the TIMP-1 assaywas 0.08ng/ml, and the standard curve ranged from020 ng/ml.Data represented in Fig. 8 are the meanSD from 4 separateexperiments done in duplicate.

4.9. MMP zymography

Commercially prepared 416% zymogen gels containinggelatin and pre-stained with blue casein (Invitrogen Corpora-tion, Carlsbad, CA) are used. M-PER protein extractionreagent (Pierce, Inc, Rockford, IL) is used to extract the proteinfrom approximately 1106 BMVEC. Typically, 30 g protein

B R A I N R E S E A R C Hisolated from untransfected and QDsiRNAMMP-9 transfectedBMVEC is loaded on to the gel 48 h post-transfection.Electrophoresis using a constant voltage of 125 V and a runtime of 60120 min in 1 Tris-glycine SDS running buffer isperformed. The gel is incubated in Zymogram RenaturingBuffer (Invitrogen Corp., Carlsbad, CA), with gentle agitationfor 30 min at room temperature, and then equilibrated inZymogram Developing Buffer (Invitrogen Corporation, Carls-bad, CA) at 37 C for at least 4 h. Areas of enzyme activityappear as clear bands against a dark blue background wherethe protease has digested the substrate. Appropriate molecu-lar weight markers and a MMP-9 standard are also run in allgels to identify the band of interest. The band densitiesrepresenting protease activity were determined by densito-metry using Syngene software (Syngene Inc, Fedrick, MD).Data are the meanSD of 3 separate experiments done induplicate. Statistical significance was determined usingANOVA based on comparisons between QDsiRNAMMP-9

nanoplexes and the negative control.

4.10. Immunofluorescent staining of BMVEC

Untransfected and transfected (control) BMVEC are grown to70% confluence in a Petri dish with a glass bottom. Cells arefixed for 10 min at 37 C in 2% formaldehyde, followed bypermeabilization with ice-cold 90% methanol. Cells are thenwashed in 1 phosphate buffered saline (PBS) and treatedwithMMP-9 antibody. Standard immunofluorescent staining pro-cedures were followed. Imaging was performed with aconfocal laser scanning microscope (TCS SP2 AOBS, LeicaMicrosystems Heidelberg GmbH) with a 63 oil immersionobjective lens. A HeNe 633 nm laser was applied to excite theAlexa Fluor 647; an argon ion laser was applied to excite thedye DAPI.

4.11. In-vitro BBB model

The in-vitro BBB model used in this study utilizes 2-compart-ment, 6 well culture plates with the upper compartmentseparated from the lower by a 3 m pore-size polyethyleneterephthalate (PET) insert with a surface area=4.67 cm2 (Trans-well plates, Costar). Untransfected and transfected BMVEC arecultured to confluence on the upper side of the PET insert, whilea confluent layer of normal human astrocytes (NHAs) are grownon the underside (Persidsky and Gendelman, 1997). Thetransendothelial electrical resistance (TEER) across the in vitroBBB isused toevaluate its functional integrity.TEER ismeasuredwith an ohm meter (Millicell ERS system, Millipore, Bedford,MA). Electrodes are sterilized using 95% alcohol and rinsed insterile distilled water prior to measurement. Probes arepositioned such that one end is immersed in media inside theupper filter chamber and the other is placed through thebasolateral access hole into the media below. A constantdistance of 0.6 cm is maintained between the electrodes at alltimes during TEER measurement. Formation of a functionallyintact BBB with untreated BMVEC occurs by day 56 of cultureand a value of 250 /cm2 indicates an intact BBB.

4.12. Statistical analysis

1532 ( 2 0 0 9 ) 1 4 2 1 5 5An independent t-test or one-way ANOVA was applied tocompare changes in different groups. Comparison among the

Brew, K., Dinakarpandian, D., Nagase, H., 2000. Tissue inhibitors of

metalloproteinase-2 in bronchopulmonary and breast lesions.

1 2 8J. Histochem. Cytochem. 47, 15751580.Current Protocols in Immunology, Copyright 2007 by JohnWiley

and Sons, Inc, Hoboken, NJ. 2007.Ellis, R., Langford, D., Masliah, E., 2007. HIV and antiretroviral

therapy in the brain: neuronal injury and repair. Nat. Rev.Neurosci. 8, 3344.

Foda, H.D., Zucker, S., 2001. Matrix metalloproteinases in cancermetalloproteinases: evolution, structure and function.Biochim. Biophys. Acta 1477 (12), 267283 Mar 7.

Burggraf, D., Liebetrau, M., Martens, H., Wunderlich, N., Jger, G.,Dichgans, M., Hamann, G.F., 2005. Matrix metalloproteinaseinduction by EMMPRIN in experimental focal cerebralischemia. Eur. J. Neurosci. 22, 273277.

Bustin, S.A., 2002. Quantification of mRNA using real-time reversetranscription PCR (RT-PCR): trends and problems. J. Mol.Endocrinol. 1, 2339.

Caudroy, S., Polette, M., Tournier, J.M., Burlet, H., Toole, B., Zucker,S., Birembaut, P., 1999. Expression of the extracellular matrixmetalloproteinase inducer (EMMPRIN) and the matrixmeans was performed with the post hoc Bonferroni analysistest using the Prism 4 statistical software (GraphPad Software,Inc. La Jolla, CA) statistical analysis software. The quantitativedata are presented as meanSD unless specified otherwise.Statistical significance was considered at p

Sundstrom, J.B., Mosunjac, M., Martinson, D.E., Bostik, P., Donahoe,R.M., Gavanis, A.A., 2001. Effects of norepinephrine, HIV type 1infection, and leukocyte interactions with endothelial cells onthe expression of matrix metalloproteinases. Ansari AIDS Res.Hum. Retrovir. 17, 16051614.

Teichberg, V.I., 2007. From the liver to the brain across thebloodbrain barrier. Proc. Natl. Acad. Sci. 104, 73157316.

Timpl, R., Wiedemann, H., van Delden, V., Furthmayr, H., Khn, K.,1981. A network model for the organization of type IV ofcollagen molecules in basement membranes. Eur. J. Biochem.12, 203211.

Wang, T., Zhang, X., Li, J.J., 2002. The role of NF-kappaB in theregulation of cell stress responses. Int. Immunopharmacol. 2(11), 15091520 Oct.

Whitehead, K.A., Langer, R., Anderson, D., 2009. Knocking downbarriers: advances in siRNA delivery. Nat. Rev. Drug Discov. 8,129138.

Woessner Jr, F., 1991. Matrix metalloproteinases and theirinhibitors in connective tissue remodeling. FASEB J. 5,21452154.

Wu, D.T., Woodman, S.E., Weiss, J.M., Mc, M.C.M., D'Aversa, T.G.,Hesselgesser, J., Major, E.O., Nath, A., Berman, J.W., 2000.Mechanisms of leukocyte trafficking into the CNS.J. Neurovirol. Suppl. 1, S82.

Yong, K.T., Qian, J., Roy, I., Lee, H., Bergey, E., Tramposch, K., He, S.,Swihart, M., Maitra, A., Prasad, P., 2007. Quantum rodbioconjugates as targeted probes for confocal and two-photonfluorescence imaging of cancer cells. Nano Lett. 7, 761765.

155B R A I N R E S E A R C H 1 2 8 2 ( 2 0 0 9 ) 1 4 2 1 5 5

MMP-9 gene silencing by a quantum dotsiRNA nanoplex delivery to maintain the integrity of the .....IntroductionResultsComplex formation of siRNA with the QDs (nanoplexes)Size and spectral characterizationCellular uptake of fluorescent siRNA after QD-mediated deliveryEffect of nanoplexes on cell viabilityEfficiency of QDsiRNAMMP-9 nanoplexes on knockdown of MMP-9 gene expression by BMVECDecreased MMP-9 enzyme activity in BMVEC transfected with QDsiRNAMMP-9 nanoplexesEffect of MMP-9 gene silencing on the expression of TIMP-1 and EMMPRIN genesEffect of MMP-9 gene silencing on MMP-9 and TIMP-1 protein expressionEffect of MMP-9 silencing on the gene and protein expression of Collagens I1, III, IV and V by.....Effect of MMP-9 gene silencing on the functional integrity of an in-vitro BBB model

DiscussionExperimental proceduresNanoparticle formulationPhysical characterization of QDCell cultureCell viability assayDetermination of the delivery of nanoplexes into BMVECReal-time, quantitative PCR (Q-PCR)Western blot analysisMeasurement of MMP-9 and TIMP-1 protein expression using ELISAMMP zymographyImmunofluorescent staining of BMVECIn-vitro BBB modelStatistical analysis

AcknowledgmentsReferences