The association of silicon microparticles with endothelial cells in drugdelivery to the vasculature

Rita E. Serda a,*, Jianhua Gu a, Rohan C. Bhavane a, XueWu Liu a, Ciro Chiappini b, Paolo Decuzzi c,d,Mauro Ferrari a,e, f,*

a Brown Institute of Molecular Medicine, University of Texas Health Science Center, Nanomedicine Division, 1825 Pressler, Suite 537, Houston, TX 77030, USAb University of Texas at Austin, Department of Biomedical Engineering, 1 University Station, C0400, Austin, TX 78712, USAc Universita’ degli Studi Magna Græcia, Viale Europa, Italyd School of Health Information Sciences, University of Texas Health Science Center Houston, 7000 Fannin, Houston, TX 77030, USAe Rice University, Department of Bioengineering, Houston, TX 77005, USAf University of Texas MD Anderson Cancer Center, Department of Therapeutics, Unit 422, 1515 Holcombe Boulevard, Houston, TX 77030, USA

a r t i c l e i n f o

Article history:Received 3 December 2008Accepted 3 January 2009Available online 12 February 2009

Keywords:EndotheliaPhagocytosisSiliconMicroparticles

a b s t r a c t

Endothelial targeting is an approach evolving for drug delivery to the vasculature of pathological lesions.Nano-porous silicon-based multi-functional particles are of particular interest, since they can be man-ufactured in essentially any size and shape, employing methods of photolithography, to optimize theirability to localize on target endothelia. In this study we tested the impact of surface charge, serumopsonization, and inflammation on the ability of vascular endothelial cells to associate with nano-poroussilicon microparticles. Vascular endothelial cells were capable of rapidly internalizing both positive andnegative silicon microparticles by an actin-dependent mechanism involving both phagocytosis andmacropinocytosis. However, following serum opsonization, internalization was selective for APTES(originally positive) modified microparticles, despite the finding that all opsonized microparticles hada net negative charge. Conversely, macrophages displayed a preference for internalization of serumopsonized oxidized (originally negative) microparticles, supporting the choice of positive microparticlesfor endothelial targeting. The internalization of opsonized microparticles by endothelial cells was furtherenhanced by the presence of inflammatory cytokines. These findings suggest that it may be possible tobioengineer silicon microparticles to favor opsonization with proteins that enhance uptake by endo-thelial cells, without a concurrent enhanced uptake by macrophages.

Published by Elsevier Ltd.

1. Introduction

Nano-structured delivery systems are emerging as powerfultools for the systemic delivery of therapeutic molecules andimaging agents for different biomedical applications, fromcancer [1,2] to cardiovascular diseases [3]. These deliverysystems can be loaded with a multitude of drug molecules andcontrast agents to simultaneously provide therapeutic andimaging capabilities. Following intravenous injection, theseparticles are transported by the blood stream into differentvascular districts. Targeting diseased vasculature is a strategyalready demonstrated to be effective for the detection of

atherosclerotic plaque and cardiovascular alterations [4,5], and itis becoming more prominent in cancer applications [6] followingthe pioneering work of Hood and co-workers [7] that introducedthe notion of targeting via integrins such as aV-b3. Typically,delivery vectors are conjugated with ligand molecules thatspecifically recognize and interact with receptors uniquely orover-expressed at pathological lesions [8–10]. Targeting ligandsinclude peptides identified from phage-displayed libraries [11],thioaptamers [12], and antibodies. Receptors for these ligandsact as vascular docking sites for the circulating particles.

The most prominent bottle-neck to systemic delivery of parti-cles, regardless of whether they are designed for vascular endo-thelial or tumor cell surface targeting, is their uptake byprofessional phagocytes, either circulating as monocytes/macro-phages or accumulating in specific organs, such as the liver, spleenand lungs. Attempts at bypassing phagocytic uptake by shieldingwith polymers, such as poly(ethylene glycol), successfully prolongscirculation time, but also limits attachment of targeting ligands anduptake by target populations [13].

* Correspondence at: Mauro Ferrari and Rita E. Serda, Brown Institute ofMolecular Medicine, Nanomedicine Division, University of Texas Health ScienceCenter, 1825 Pressler Street, Suite 537, Houston, TX 77030, USA. Tel.: þ1 713 5002309; fax: þ1 713 500 2313.

E-mail addresses: rita.serda@uth.tmc.edu (R.E. Serda), mauro.ferrari@uth.tmc.edu (M. Ferrari).

Contents lists available at ScienceDirect

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

0142-9612/$ – see front matter Published by Elsevier Ltd.doi:10.1016/j.biomaterials.2009.01.019

Biomaterials 30 (2009) 2440–2448

Phagocytosis of particulates, apoptotic bodies, and aged eryth-rocytes is predominately attributed to macrophages, neutrophilsand dendritic cells. However, several studies have describedphagocytosis of large objects by endothelial cells, including theuptake of latex particles, zymosan, apoptotic bodies and microor-ganisms [14–18]. In this study, the role of endothelial cells incellular uptake of porous silicon microparticles is explored. Theadvantages of using porous silicon for drug delivery includebiocompatibility, exquisite control over rates of degradation, andprecise control over size and shape by employing methods ofphotolithography [2,19,20]. Precise control over particle geometryallows realization of the computational prediction that size andshape influence vascular navigation, avoidance of biologicalbarriers, localization, and biological pathways [21–23]. The porousnature of the silicon microparticles transforms the conceptual ideaof multi-functional particles into a reality by creating vectorscapable of releasing successive stages of carries and drugs intendedto conquer sequential biological barriers to reach the intendedtarget [24]. In this study, the effect of surface chemistry, serumopsonization and inflammation on the interplay between siliconmicroparticles and endothelial cells is explored.

2. Materials and methods

2.1. Silicon particle fabrication

Nano-porous hemispherical silicon microparticles were designed, engineered,and fabricated in the Microelectronics Research Center at The University of Texas atAustin. Two sizes of microparticles were generated, with mean diameters of1.6" 0.2 and 3.2" 0.2 mm, and pore sizes ranging from 5 to 10 nm. Processing detailswere recently published by our laboratory [2]. Briefly, heavily doped pþþ type (100)silicon wafers with resistivity of 0.005 ohm cm (Silicon Quest, Inc., Santa Clara, CA)were used as the silicon source. A 100 nm layer of low-stress silicon nitride wasdeposited using a Low Pressure Chemical Vapor Deposition (LPCVD) system. Stan-dard photolithography was used to pattern the microparticles over the wafer usinga contact aligner (EVG 620 aligner) and AZ5209 photoresist. Nitride on particlepatterns was selectively removed by CF4 based reactive ion etching (RIE). After thephotoresist was stripped in piranha solution, the wafer was placed in a home-madeTeflon cell for two-step electrochemical etching. First, the wafers were etched ina mixture of hydrofluoric acid (HF) and ethanol (1:1 v/v) by applying a currentdensity of 6 mA/cm2 for 105 s for 3.2 mm particles or 40 s for 1.6 mm particles,respectively. Then a high porosity release layer was formed by changing the currentdensity to 320 mA/cm2 for 6 s in a 2:5 v/v mixture of HF and ethanol. Finally, thenitride layer was removed in HF after etching, and microparticles were released byultrasound in isopropyl alcohol (IPA) for 1 min. The IPA solution containing poroussilicon microparticles was collected and stored at 4 #C. The morphology of themicroparticles was examined by SEM.

2.2. Oxidation of silicon microparticles

Silicon microparticles in isopropyl alcohol (IPA) were dried in a glass beaker kepton a hot plate (110 #C). The dried microparticles were then treated with piranhasolution (1 volume H2O2 and 2 volumes of H2SO4). The suspension was heated to110–120 #C for 2 h with intermittent sonication to disperse the microparticles. Thesuspension was then washed in deionized (DI) water until the pH of the suspensionwas w5.5–6.

2.3. Surface modification of silicon microparticles with APTES

The oxidized microparticles were washed in IPA 3–4 times. They were thensuspended in IPA containing 0.5% (v/v) APTES (Sigma) for 2 h at room temperature.The APTES modified microparticles were washed and stored in IPA. APTES modifi-cation was evaluated by measuring the zeta potential and by colorimetric analysis ofamine density. The latter was found to correlate with zeta potential measurements.

2.4. PEG conjugation

APTES modified microparticles were reacted with 10 mM mPEG-SCM-5000(methoxy poly-ethylene glycol succinimidyl carboxymethyl; purchased from LaysanBio Inc.) in acetonitrile for 1.5 h. The microparticles were then washed in distilledwater 4–6 times to remove any unreacted mPEG. Zeta potential measurements wereused to indicate adequate surface coating.

2.5. Z2 analysis

Microparticles were counted in a Z2 Coulter! Particle Counter and Size Analyzer(Beckman Coulter). The aperture size used for microparticle analysis was 50 mm. Foranalysis, microparticles were suspended in the balanced electrolyte solution of theinstrument (ISOTON! II Diluent) and counted.

2.6. Cell culture

HUVEC, purchased from Lonza Walkersville, Inc. (Walkersville, Maryland), werecultured in EBM!-2 medium (Clonetics!, CC-3156). Cells were maintained at 37 #Cin a humidified 5% CO2 atmosphere, and detached using a 0.25 mg/ml trypsin/EDTAsolution (Clonetics!) upon reaching 80–90% confluency. For serum-free experi-ments, the media used was EBM!-2 medium (Clonetics!) supplemented with onlyhydrocortisone and GA-1000, plus 0.2% BSA. HUVECs were discarded after 7–8passages. J774A.1 macrophage cells were purchased from American Type CultureCollection (Manassas, VA). Growth medium was Dulbecco’s Modified Eagle’sMedium containing 10% FBS, 100 mg/ml streptomycin and 100 U/ml Penicillin(Invitrogen; Carlsbad, CA). Cells were collected by scrapping.

2.7. Confocal microscopy

HUVECs were grown on No. 1.5 glass coverslips. When confluent, cells wereincubated with microparticles (1:10; cell:microparticle) for 60 min in serum-freemedia at 37 #C. Cells were then washed with PBS, fixed and premeabilized with 0.1%triton X-100. PBS containing 1% BSA was used as a blocking agent prior to incubationwith 200 nM Alexa Fluor 555 conjugated phallodin (Invitrogen) and anti-tubulinFITC conjugated antibody (Abcam Inc., Cambridge, MA). Coverslips were thenwashed and mounted on glass slides using Vectashield mounting media (VectorLaboratories, Burlingame, CA) containing a 1000-fold dilution DRAQ5 (BiostatusLimited, UK). Detection of silicon microparticles was based on autofluorescenceusing a 633 excitation laser. Images were acquired using a Leica DM6000 uprightconfocal microscope equipped with a 63$ oil immersion objective.

2.8. Flow cytometry

HUVECs (1.5$105 cells/well) were seeded into 6 well plates and 24 h later thecells were incubated with silicon microparticles (10–20 microparticles/cell) inserum-free media, or as specified, for the indicated time. Fluoresbrite! YG Micro-spheres (1 mm; Polysciences) were incubated with HUVECs at 37 #C for 60 min ata ratio of 1:1000 (cell:particle). For opsonization experiments, microparticles werepre-incubated with 100 ml of 100% FBS or physiological concentrations of BSA (4 g/dLin PBS) or ImmunoPure! Human Whole Molecule IgG (500 mg/dL; Pierce, Rockford,IL) on ice for 60 min. Following incubation with microparticles, cells were washedwith PBS, harvested by trypsinization (HUVEC) or scrapping (J774), and resuspendedin PBS containing 1.0% BSA and 0.1% sodium azide (FACS wash buffer). Microparticleassociation with cells was determined by measuring side scatter using a BectonDickinson FACSCalibur Flow equipped with a 488-nm argon laser and CellQuestsoftware (Becton Dickinson; San Jose, CA). Data is presented as either percentage ofcells with microparticles (percent of cells with high side scatter) or mean particleuptake per cell (mean side scatter). Side scatter due to cells in the absence ofparticles has been subtracted from the presented data. Comparisons of FITC Dextran(0.5 mg/ml) internalization by HUVECs are based on fluorescent intensitymeasurements following incubation of cells with microparticles at 37 #C for 1 h.

The following antibodies were used to measure expression of Fc gamma globulinreceptors: Monoclonal anti-Human CD64 FITC Conjugate (clone 10.1, BD Pharmin-gen), Monoclonal anti-Human CD32 FITC Conjugate (clone 3D3; BD Pharmingen),and Monoclonal anti-Human CD16 FITC Conjugate (clone 3G8; Sigma). Expression ofFcgR, as well as microparticle uptake, were also studied on HUVECs stimulated for48 h with TNF-a (10 ng/ml) and IFN-g (100 U/ml). Flow cytometric analysis wasperformed on live cells, which were identified by their lack of propidium iodidestaining. Antigen density calculations were based on a QuickCal! calibration curvegenerated with Quantum" Simply Cellular! anti-Mouse IgG (Bangs Laboratories,Inc.; Fishers, IN).

2.9. Inhibition studies

A 5 mg/ml Cytochalasin B solution in 95% ethanol (Sigma–Aldrich; St. Louis, MO)was prepared and stored at %20 #C. For inhibition studies, HUVEC cells, plated in 6well plates (80–90% confluent), were incubated in media containing Cytochalasin B(2.5, 5.0 and 10.0 mg/ml) for 60 min at 37 #C prior to addition of microparticles andduring the 60 min incubation with microparticles (4$106 microparticles per well;ratio 1:20). Following incubation, the cells were washed with PBS, collected bytrypsinization and resuspended in FACS wash buffer. Samples were analyzed by flowcytometric analysis.

R.E. Serda et al. / Biomaterials 30 (2009) 2440–2448 2441

2.10. Scanning electron microscopy

Microparticle binding and uptake by HUVEC and J774 cells were observed byscanning electron microscopy. Cells were plated in 24 well plates containing5$7 mm Silicon Chip Specimen Supports (Ted Pella, Inc., Redding, CA) at 5$104

cells per well. When cells were confluent, media containing microparticles (1:20,cell:microparticles, 0.5 ml/well) was introduced and cells were incubated at 37 #Cfor the specified amount of time. Samples were washed with PBS and fixed in 2.5%glutaraldehyde for 30 min (Sigma–Aldrich; St. Louis, MO). After washing in PBS, cellswere dehydrated in ascending concentrations of ethanol (30%, 50%, 70%, 90%, 95%,and 100%) for 10 min each. HUVECs were then incubated in a 50% alcohol–hexamethyldisilazane (HMDS; Sigma) solution for 10 min followed by incubation in pureHMDS for 5 min and overnight incubation in a desiccator. Specimens were mountedon SEM stubs (Ted Pella, Inc.) using conductive adhesive tape (12 mm OD PELCOTabs, Ted Pella, Inc.). Samples were sputter coated with a 10 nm layer of gold usinga Plasma Sciences CrC-150 Sputtering System (Torr International, Inc.). SEM imageswere acquired under high vacuum, at 20.00 kV, spot size 3.0–5.0, using an FEIQuanta 400 FEG ESEM equipped with an ETD (SE) detector.

2.11. Transmission electron microscopy

HUVECs were grown to 80% confluency in a 6 well plate. Using 1 ml of serum-free media per well, 1.6 mm or 3.2 mm particles were introduced at a cell:micro-particle ratio of 1:10 at 37 #C for either 15 or 120 min. HUVECs were then washedand fixed in a solution of 2% paraformaldehyde (16% solution purchased fromElectron Microscopy Sciences; Hatfield, PA) and 3% glutaraldehyde (25% solutionfrom Sigma) in phosphate buffered saline pH 7.4 (2 ml/well; Sigma) for 1 h. Afterfixation, the samples were washed and treated with 0.1% Millipore-filtered caco-dylate buffered tannic acid, post-fixed with 1% buffered osmium tetroxide for30 min, and stained en bloc with 1% Millipore-filtered uranyl acetate. The sampleswere dehydrated in increasing concentrations of ethanol, infiltrated, and embeddedin Poly-bed 812 medium. The samples were polymerized in a 60 #C oven for 2 days.Ultrathin sections were cut using a Leica Ultracut microtome (Leica, Deerfield, IL),stained with uranyl acetate and lead citrate in a Leica EM Stainer, and examined ina JEM 1010 transmission electron microscope (JOEL, USA, Inc., Peabody, MA) at anaccelerating voltage of 80 kV. Digital images were obtained using the AMT ImagingSystem (Advanced Microscopy Techniques Cory, Danvers, MA).

0 µm 5

2µm

5µm

A

B

30 min 60 min

cells

5 min 60 min

15 min

Fig. 1. Internalization of silicon microparticles by endothelial cells. (A) HUVECs, grown on silicon supports, were incubated with 3.2 mm oxidized silicon microparticles (10:1particles per cell) for 15–60 min at 37 #C in serum-free media, followed by fixation, dehydration, and imaging by SEM. (B) Confocal micrographs of HUVECs incubated with 3.2 mmoxidized silicon microparticles (10:1) for 5 and 60 min at 37 #C. Following incubation, cells were fixed, permeabilized, and stained with Alexa Fluor 555 Phalloidin. Microparticleautofluoresence was imaged following excitation at 633 nm.

R.E. Serda et al. / Biomaterials 30 (2009) 2440–24482442

3. Results

3.1. Silicon microparticle uptake by endothelial cells

Nano-structured silicon microparticles were fabricated usinga porosification process involving photolithography [25]. Thisprocess allows for accurate control over particle size and geometry,two parameters important for drug delivery. To examine theresponse of endothelial cells to silicon microparticles, two sizes ofmicroparticles, 1.6" 0.2 and 3.2" 0.2 mm, and three types ofsurface modifications were explored. All experiments were per-formed with both sizes of microparticles in triplicate, and repre-sentative experiments are presented.

Initial experiments explored the response of endothelial cells tooxidized silicon microparticles. Microparticles were introduced toHuman Umbilical Vein Endothelial Cells (HUVECs) at a ratio of 20particles to every cell. Scanning electron microscopy (SEM) imagesdemonstrated that early contact (15 min at 37 #C) of HUVECs withmicroparticles induced formation of pseudopodia that extendedoutward to engulf the microparticles (Fig. 1A). SEM analysis at30–60 min showed an increasing number of microparticlesbecoming completely internalized by HUVECs with time. Confocalmicroscopy images, taken after 5 min of incubation of microparti-cles with HUVECs at 37 #C, verified the formation of cellularmembrane extensions reaching out to surround the microparticles(Fig. 1B). After 60 min, the internalized microparticles were locatedin the perinuclear region of the cell. Microparticles were visualizedby autofluorescence (excitation at 633 nm) and the cell periphery(i.e. actin) was labeled with Alexa Fluor 555 conjugated phalloidin.Based on an MTT cell proliferation assay, at 12–72 h cell viabilitywas similar for cells cultured in media alone versus media con-taining both 5 and 10 particles to every cell (Appendix S1).

3.2. Size effect on microparticle uptake

Both phagocytosis and macropinocytosis are characterized byextensive membrane reorganization. Phagocytic engulfment ischaracterized by the formation of tight-fitting phagosomes, whilemacropinocytotic uptake involves the formation of ruffles thatengulf both solid cargo and fluid [26]. In this study, HUVECs incu-bated with 1.6 mm silicon microparticles for 15 min at 37 #C wereengulfed by mechanisms phenotypically resembling both macro-pinocytosis and phagocytosis (Fig. 2A). Macropinocytosis ofparticles by endothelial cells is supported by Hartig et al. [27] whorecently reported that nonspecifically bound nanoparticulatecomplexes are taken up by microvascular endothelial cells bymacropinocytosis. Internalization of the larger 3.2 mm particles byHUVECs was more consistent with classical phagocytosis, with thehallmark presence of an actin cup holding a microparticle (Fig. 2B).

Involvement of macropinocytosis in microparticle uptake wassupported by internalization of fluorescein isothiocyanate dextran(FD) from the cell media during microparticle uptake. A greateramount of FD was internalized by HUVECs in the presence of 1.6 mmparticles compared to 3.2 mm particles (Fig. 2C) supportinga greater role for macropinocytosis in the uptake of the smallermicroparticles.

3.3. Effect of Cytochalasin B on silicon microparticle uptake

Membrane reorganization occurring during phagocytosis andmacropinocytosis involves reorganization of the actin cytoskel-eton, which leads to changes in the shape of the cell membrane.Cytochalasin B dissociates actin-binding protein from actin fila-ments, weakening existing actin filaments, and blocking actinpolymerization [28]. To confirm the involvement of actin

Fig. 2. Early uptake of oxidized silicon microparticles by endothelial cells. HUVECswere incubated with either 1.6 mm (A) or 3.2 mm (B) oxidized silicon microparticles(10:1 particles per cell) for 15 min at 37 #C in serum-free media. HUVECs were washed,fixed and processed for TEM analysis. Top row: direct magnification 3000$; others25,000$. (C) Internalization of FITC dextran, as measured by flow cytometry, was usedas a metric to measure fluid uptake indicative of macropinocytosis. HUVECs wereincubated with either 1.6 mm (red line) or 3.2 mm (purple line) oxidized siliconmicroparticles in the presence of FITC dextran for 60 min. Controls included cellsincubated with FITC dextran (green peak) or media (blue peak). (For interpretation ofthe references to colour in this figure legend, the reader is referred to the web versionof this article.)

R.E. Serda et al. / Biomaterials 30 (2009) 2440–2448 2443

polymerization in the uptake of silicon microparticles, HUVECswere pre-incubated with Cytochalasin B (2.5 mg/ml) for 60 min,then incubated with both 1.6 mm and 3.2 mm oxidized micropar-ticles for 15–60 min at 37 #C. SEM analysis at 15 min shows a lackof pseudopodia engaging microparticles which have settled on thecell surface (Fig. 3A). After 30 min, the HUVECs appear to haveinitiated actin cup formation, which is a dense actin network thatforms beneath the microparticles [29]; however, uptake is frus-trated by Cytochalasin B, preventing internalization of themicroparticles. In the absence of Cytochalasin B, further actinpolymerization would induce expansion of the actin cup whichwould completely surround the microparticle, creating a phag-osome. Confocal images taken after 60 min of incubation showmicroparticle uptake by HUVECs in the absence of Cytochalasin B(control), however, in agreement with SEM images, uptake ofmicroparticles is blocked by addition of Cytochalasin B (CB)(Fig. 3B). Filipin and chlorpromazine, inhibitors of caveolae andclathrin mediated endocytosis, failed to block the uptake of siliconmicroparticles by HUVECs as visualized by confocal microscopy(R.E. Serda, unpublished data).

Cells containing internalized microparticles have an increase ingranularity as demonstrated by an upward shift in orthogonal (90#)laser light scatter (side scatter) based on flow cytometric analysis. Asthe ratio of microparticles to cells increases, there is a linear increasein the percentage of cells with a high side scatter phenotype (Fig. 3C).Based on gating similar to that shown in Fig. 3D, side scatter was usedas a metric to measure microparticle uptake by HUVECs (i.e. percentcells with internalized microparticles). Dose dependent inhibition ofmicroparticle uptake by HUVECs was evaluated by flow cytometry atconcentrations of Cytochalasin B of 2.5 mg/ml and higher. Cytocha-lasin B inhibited microparticle uptake by 85–88% (Fig. 3E).

3.4. Effect of surface charge on microparticle uptake

The effect of surface charge on microparticle uptake was eval-uated by flow cytometry and SEM analysis. Silicon microparticleswere oxidized with a piranha solution [30:70 (v/v) H2O2:H2SO4] tocreate negatively charged, hydroxylated microparticles. Completesurface oxidation was confirmed by energy dispersive spectro-photometer analysis (data not shown). Surface modification with

2500 µm 25 µm

25 µm

HV20.00 kV

spot5.5

mag7.500 x

WD9.2 mm

HFW19.9 µm

5 µmHUVEC 1X

6/28/200711:22:32 AM

HV20.00 kV

spot5.5

mag30.000 x

WD9.7 mm

HFW4.97 µm

1 µmHUVEC 1X

7/8/200711:38:15 AM

y = 15.4x + 0.4R2 = 0.9605

0102030405060708090

0 5 10 20 40particle to cell ratio

pa

rtic

le u

pta

ke

02468

10121416

cells

% c

ell

s w

ith

pa

rtic

les

cells

Cytochalasin B (CB)

Dcontrol60 min

C

cells +Particles(MP)

CB15 min

CB30 min

CB60 min

E

0 200 400 600 800 1000 0 200 400 600 800 1000Forward Scatter Forward Scatter

1000

600

800

400

200

0S

ide

Sc

atte

r

1000

600

800

400

200

0S

ide

Sc

atte

r

MP CB 2.5 CB 10 CB 25

A

B

Fig. 3. Cytochalasin B blocks microparticle uptake by HUVECs. (A) SEM images of HUVECs incubated with a mixture of 1.6 mm and 3.2 mm oxidized silicon microparticles at 37 #C for15 and 30 min in serum-free media containing 2.5 mg/ml Cytochalasin B. (B) Confocal micrographs of control or Cytochalasin B treated HUVECs incubated with 3.2 mm siliconmicroparticles for 60 min and stained with FITC conjugated anti-tubulin antibody, Alexa Fluor 555 Phalloidin (actin), and DRAQ5 (nuclear). (C) Flow cytometric analysis showingorthogonal 90# laser light scatter (side scatter) by HUVECs after incubation with increasing numbers of microparticles (0–40 particles per cell) for 60 min at 37 #C. (D) Flowcytometry dot blots showing the increase in side scatter caused by cell uptake of microparticles. Control cells are shown on the left and HUVECs incubated with 10 microparticlesper cell for 60 min (37 #C) are to the right. The box designates cells with internalized microparticles (MP), which is illustrated in the graph below. (E) HUVECs, pretreated for 60 minwith 2.5–25 mg/ml Cytochalasin B (CB), were incubated with silicon microparticles for an additional 60 min at 37 #C. Cells were harvested by trypsinization and particle uptake (i.e.side light scatter) was measured as shown in ‘‘D’’.

R.E. Serda et al. / Biomaterials 30 (2009) 2440–24482444

3-aminopropyltriethoxysilane (APTES) yielded positively charged,amine modified microparticles. APTES modified microparticleswere further conjugated with methoxy poly-ethylene glycol-5000(mPEG-5000) to create shielded microparticles with a more neutralcharge.

In the presence of serum-free media, HUVECs non-discriminately internalized both positive and negative microparti-cles (Fig. 4A). SEM images are of cells after 1 h at 37 #C withmicroparticles. Since it is well established that PEG creates a shieldaround particles, causing them to avoid or delay phagocytosis bymacrophages, we examined uptake of mPEG-5000 microparticlesby endothelial cells. Surface modification of silicon microparticleswith mPEG-5000 effectively suppressed internalization of micro-particles by HUVECs at 1 h (Fig. 4A).

3.5. Effect of serum opsonization on microparticle uptake

Microparticles were incubated with 100% fetal bovine serum(hereafter serum) for 60 min on ice, followed by incubation with

HUVECs in the continued presence of 5% serum. Based on zetapotential measurements, despite the original surface charge, allopsonized silicon microparticles had a net negative surface charge(Fig. 4B). Based on flow cytometric analysis, while internalization ofoxidized microparticles by HUVECs was strongly inhibited (78% lessuptake than non-opsonized oxidized microparticles), serum opso-nization failed to have an impact on internalization of APTESmicroparticles by HUVECs. Opsonization of PEG modified micro-particles lead to a 27% decrease in uptake by endothelial cells(Fig. 4C). To evaluate the generality of inhibition of uptake ofnegatively charged microparticles by endothelial cells followingserum opsonization, negatively charged polystyrene microparticles(Fluoresbrite! YG, spherical, 1 mm; Polysciences) were opsonized inserum, and then incubated with HUVECs in media containing 5%serum. In the presence of serum, there was a 33-fold decrease inuptake of negative polystyrene microparticles, supporting theconclusion that serum blockade of internalization of originallynegatively charged microparticles is a global phenomenonregardless of the type of microparticle used (Appendix S2).

010203040506070

oxidized APTES PEG

pa

rtic

le u

pta

ke

serum freeserum

0

10

20

30

40

50

oxidized APTES

pa

rtic

le u

pta

ke

serum-freeserum

serum-free serum

OXID -43.48 ± 1.65 -32.21 ± 3.43

APTES 15.12 ± 4.33 -40.28 ± 2.78

PEG -3.84 ± 1.99 -27.82 ± 1.71

oxid

ized

APTE

SPE

G

Endothelial cells - serum-free

*

Macrophage

***

Endothelial

AB

C

D

Fig. 4. Silicon microparticle surface charge and binding by serum opsonins directs uptake by endothelial cells. (A) HUVECs grown on silicon supports were incubated with non-serum opsonized negative (oxidized), positive (APTES), or mPEG-5000 modified microparticles for 60 min at 37 #C. SEM images were taken at direct magnification 10,000$ (left)and 20,000$ (right). (B) Electrostatic (zeta) potential of microparticles before and after serum opsonization (60 min, 4 #C). (C) Flow cytometric analysis of microparticle uptake byHUVECs (20:1; microparticles:cell) in serum-free or serum-containing media (*p< 0.0004; **p< 0.015). Particle uptake is the percent of cells with high side scatter. (D) Flowcytometry analysis of microparticle uptake by J774 macrophages in serum-free or serum-containing media (*p< 0.03).

R.E. Serda et al. / Biomaterials 30 (2009) 2440–2448 2445

Similar to endothelial cells, J774 mouse macrophages were ableto take up positive and negative silicon microparticles non-discriminately in serum-free media (Fig. 4D). However, undernormalized conditions with serum present, J774 cells were alsoresponsive to signals provided by particle bound opsonins. Incontrast to endothelial cells, macrophages favored uptake ofopsonized, originally negatively charged, oxidized microparticlesby 29% (p< 0.05).

Opsonization of particulates by antibodies enhances phagocy-tosis by macrophages via binding to gamma immunoglobulin (IgG)receptors (FcgRs) on the cell surface [30]. However, FcgRs arereported to be either absent or expressed at low levels on unsti-mulated endothelial cells [31–33]. IgG opsonization would there-fore be expected to inhibit uptake of particles by endothelial cells.Additionally, Arvidsson et al. [34] has shown rapid (1 min) bindingof antibody [specific for high molecular weight kininogen (HMWK)]to oxidized silicon immersed in plasma. As expected, opsonizationof silicon microparticles in ImmunoPure Human Whole MoleculeIgG at 500 mg/dL IgG (normal range, 400–1800 mg/dL) inhibitedthe uptake of oxidized microparticles by HUVECs, both in antibodyfree media (88%) and in media supplemented with antibody (76%)(Fig. 5A). Consistent with published data, IgG opsonizationenhanced microparticle uptake by macrophages (by 35% and 47% inthe absence or presence of free IgG, respectively) (Fig. 5B). Incontrast to negative microparticles, IgG opsonization of positivemicroparticles failed to have an impact on microparticle uptake byendothelial cells, suggesting either inadequate binding or anincorrect orientation by bound IgG (Appendix S3).

Leroux et al. [35] reported that opsonization of PEG-coatednanoparticles by serum proteins enhanced their uptake by mono-cytes. They further showed that IgG present in the serum wasresponsible for the enhanced uptake of PEGylated particles. Here

we show that serum opsonization of PEGylated microparticlesdecreased their uptake by endothelial cells (Fig. 4C), consistentwith blocked uptake of IgG-opsonized negative microparticles byendothelial cells.

Analysis of FcgR expression on cells by flow cytometry confirmedhigh levels of FcgRII expression by macrophages. On the other hand,endothelial cells lacked expression of both FcgRI and FcgRIII, andexpressed low levels of FcgRII (Fig. 5C). Based on a calibration curvegenerated using Quantum" Simply Cellular! microspheres, HUVECsexpress approximately 10,000 FcgRII per cell, compared to approxi-mately 320,000 FcgRII per macrophage. Treatment of cells with TNF-a (10 ng/ml) and IFN-g (100 U/ml) for 48 h increased expression ofFcgRII on HUVECs to approximately 35,500 receptors per cell (Fig. 5D).

3.6. Phagocytosis following cytokine stimulation of HUVECs

Since it has been reported that activation of endothelial cells bypro-inflammatory cytokines enhances expression of cell adhesionmolecules [36,37], as well as FcgRs [31], microparticle uptakefollowing 48 h stimulation with TNF-a (10 ng/ml) and IFN-g (100 U/ml) was examined by flow cytometric analysis. The data isexpressed in terms of mean side scatter to give a measure of therelative number of particles per cell (scatter due to control cells hasbeen subtracted). Internalization of serum opsonized siliconmicroparticles by HUVECs was enhanced for all groups of micro-particles; however, a clear preference for opsonized APTES micro-particles continued to exist. Following cytokine stimulation,1.9-foldmore APTES microparticles were engulfed compared to oxidizedmicroparticles (Fig. 6A). J774 macrophages continued to showa clear preference for internalization of opsonized oxidized micro-particles (1.6-fold greater uptake than positive microparticles).Cytokine stimulation did not affect the number of microparticles

05

1015202530

serum-free IgG opson IgG opson +IgG media

% c

ell

s w

ith

pa

rtic

les

05

1015202530

serum-free IgG opson IgG opson +IgG media

% c

ell

s w

ith

pa

rtic

les

MacrophagesEndothelial cells

0

100000

200000

300000

400000

500000

HUVEC stimulatedHUVEC

J774 stimulatedJ774

IgG

FcγRIFcγRIIFcγRIII

re

ce

pto

rs

pe

r c

ell

A B

C DFcγγRII

9,173

35,542

324,343

313,018

HUVEC

Stim. HUVEC

J774

Stim. J774

Fig. 5. IgG opsonization blocks uptake of negative microparticles by HUVECs while stimulating uptake by J774 macrophages. Flow cytometry analysis of non-serum opsonizedversus IgG-opsonized oxidized microparticle uptake by HUVECs (A) and J774 (B) cells. Oxidized microparticles were pre-opsonized with pure IgG for 60 min on ice, then incubatedwith cells for 60 min in serum-free media at 37 #C (10 particles per cell). Cells were trypsinized following incubation and side scatter (percent cells with high SSC) was measured byflow cytometry. (C–D) Basal and stimulated [TNF-a (10 ng/ml) and IFN-g (100 U/ml) for 48 h] HUVECs and J774 cells were labeled with fluorescein isothiocyanate (FITC)-conjugatedantibodies specific for FCgRI, FCgRII, and FCgRIII. Surface expression was determined by flow cytometric analysis. Quantitative expression of receptors was determined usinga QuickCal! calibration curve. The data shown is the number of receptors expressed per cell.

R.E. Serda et al. / Biomaterials 30 (2009) 2440–24482446

engulfed by macrophages (Fig. 6B). PEGylation had a major impacton microparticle uptake by macrophages, with 2.8-fold less uptakecompared to negative, oxidized microparticles. A novel finding wasthat opsonization of negative microparticles inhibited their uptakeby endothelial cells to an extent equal to that of PEGylation, underboth basal and inflammatory conditions (Fig. 6A).

SEM images taken after 15 min at 37 #C show the membranesextending outward to wrap around serum opsonized APTESmicroparticles by HUVECs (Fig. 6C) and serum opsonized oxidizedmicroparticles by J774 cells (Fig. 6D).

4. Discussion

In this study we found that cationic silicon microparticles werepreferentially internalized by endothelial cells via phagocytosis andmacropinocytosis. These observations are consistent with recentfindings on liposomes, which suggest that cationic particles havea propensity for localizing in tumor vessels due to selectiveadherence to endothelial cells [38].

Several in vivo studies suggest that the pattern of opsoninsadsorbed to the surface of particulates determines the population ofphagocytic cells responsible for their clearance [39]. Current studiesare underway to determine which serum constituents bind topositive silicon microparticles and act as opsonins for the uptake ofmicroparticles by endothelial cells. Antibodies, which are opsoninswith respect to particle uptake by macrophages, were dys-opsoninswith respect to the uptake of microparticles by endothelial cells. Wehave shown that antibody opsonization selectively impacted theuptake of oxidized, anionic silicon microparticles. Thus, irrespectiveof size, cationic particles, ranging from nano to micron size, aretargets for binding and uptake by vascular endothelial cells.

We observed that microparticle internalization by endothelialcells was enhanced by pro-inflammatory cytokine stimulation,supporting superior uptake of silicon microparticles at sites ofchronic inflammation. At sites of pathological inflammation, up-regulation of adhesion molecules on endothelial cells leads torecruitment of leukocytes [40]. As just presented, the alteredrecognition properties of endothelial cells also inspire enhancedbinding to porous silicon microparticles. Silicon microparticleseither remained attached to the endothelial membrane or wereinternalized via phagocytosis. The array of opsonins having themost dramatic influence on the interaction between endothelia andsilicon microparticles were those binding to the surface of posi-tively charged microparticles. Future studies aimed at optimizingthe adherence of microparticles to inflamed endothelium willsimultaneously incorporate targeting ligands selective for enrichedadhesion molecules and maintenance of an overall positive chargeto favorably influence binding by endothelial specific opsonins.

Another potential targeting strategy that emerges from thisstudy takes into account microparticles that bind to the endothelialcell surface, but have not been internalized. Relying on enhancedbinding of favorable opsonized cationic microparticles at inflam-matory lesions, subsequent transmigration could be supported byattaching targeting molecules to the surface of silicon microparti-cles that are involved in secondary steps of transmigration. Forexample, initiation of transmigration involves PECAM, withcompletion depending on CD99 [41,42]. Attachment of CD99 to themicroparticle surface would thus favor migration across theendothelial barrier. Once across, the release of secondary particles,such as iron oxide nanoparticles and liposomes, from the pores ofthe silicon microparticles would home to the tumor cell itself viacancer specific targeting ligands. With this logic, one may envision

0

50

100

150

200

250

oxidized aptes peg

control

stimulated

0

50

100

150

200

250

oxidized aptes peg

control

stimulated

Macrophage

Endothelial

*

Pa

rtic

le u

pta

ke

A

B

C

D

1 micron

2 micron

Fig. 6. Inflammatory cytokines stimulate silicon microparticle uptake by endothelial cells. (A, B) Flow cytometry analysis of serum opsonized, silicon microparticle uptake byHUVECs (A) and J774 cells (B) in cells under basal conditions (control) or following stimulation with TNF-a and IFN-g for 48 h (*p< 0.045). Cells were incubated with 20 micro-particles (negative, positive, or neutral PEG) per cell for 60 min at 37 #C. Microparticle uptake is presented as mean side scatter, that is, average uptake per cell. (C, D) SEM images ofsilicon microparticle uptake by HUVEC (C) and J774 (D) cells (30 min, 37 #C) in the presence of serum.

R.E. Serda et al. / Biomaterials 30 (2009) 2440–2448 2447

transforming biological barriers into stepping stones leading to thetargeted tumor lesion.

5. Conclusions

In summary, cationic silicon microparticles bind serum opsoninsthat favor uptake by endothelia, with enhanced uptake by cytokineinflamed endothelia. Strategies for the attachment of targetingligands to the microparticle surface should incorporate methodsthat preserve or impart an overall positive charge to the micro-particle surface. The optimal design of delivery vehicles must takeinto account the final presentation of the opsonized particle.

Acknowledgements

We wish to thank Angelo Benedetto and the Rice Shared ResourceFacility for training in scanning electron microscopy and KennethDunner Jr. at MD Anderson Cancer Research Center for sampleprocessing and image analysis by transmission electron microscopy.We also thank Dr. Fredika Robertson for advice on the technicalwriting of the manuscript and Matt Landry for aid in assembling highresolution figures for publication. This research was supported bygrants DODW81XWH-07-1-0596, DODW81XWH-04-2-0035 andDODW81XWH-07-2-0101; NASA Grant #SA23-06-017; and State ofTexas, Emerging Technology Fund.

Appendix. Supplementary data

Supplementary data associated with this article can be found inthe online version, at doi:10.1016/j.biomaterials.2009.01.019.

References

[1] Ferrari M. Cancer nanotechnology: opportunities and challenges. Nat Rev2005;5:161–71.

[2] Tasciotti E, Liu X, Bhavane R, Plant K, Leonard A, Price B, et al. Mesoporoussilicon particles as a multistage delivery system for imaging and therapeuticapplications. Nat Nanotechnol 2008;3:151–7.

[3] Cyrus T, Lanza GM, Wickline SA. Molecular imaging by cardiovascular MR.J Cardiovasc Magn Reson 2007;9:827–43.

[4] Lanza G, Winter P, Cyrus T, Caruthers S, Marsh J, Hughes M, et al. Nano-medicine opportunities in cardiology. Ann N Y Acad Sci 2006;1080:451–65.

[5] Kang HW, Torres D, Wald L, Weissleder R, Bogdanov Jr AA. Targeted imaging ofhuman endothelial-specific marker in a model of adoptive cell transfer. LabInvest 2006;86:599–609.

[6] Hu G, Lijowski M, Zhang H, Partlow KC, Caruthers SD, Kiefer G, et al. Imaging ofVx-2 rabbit tumors with alpha(nu)beta3-integrin-targeted 111In nano-particles. Int J Cancer 2007;120:1951–7.

[7] Hood JD, Bednarski M, Frausto R, Guccione S, Reisfeld RA, Xiang R, et al. Tumorregression by targeted gene delivery to the neovasculature. Science2002;296:2404–7.

[8] Rossin R, Muro S, Welch MJ, Muzykantov VR, Schuster DP. In vivo imaging of64Cu-labeled polymer nanoparticles targeted to the lung endothelium. J NuclMed 2008;49:103–11.

[9] Ding B, Dziubla T, Shuvaev VV, Muro S, Muzykantov VR. Advanced drugdelivery systems that target the vascular endothelium. Adv Drug Deliv Syst2006;6:98–112.

[10] Koning GA, Schiffelers RM, Wauben MHM, Kok RJ, Mastrobattista E, Molema G,et al. Targeting of angiogenic endothelial cells at sites of inflammation bydexamethasone phosphate-containing RGD peptide liposomes inhibitsexperimental arthritis. Arthritis Rheum 2006;54:1198–208.

[11] Arap W, Haedicke W, Bernasconi M, Kain R, Rajotte D, Krajewski S, et al. Tar-geting the prostate for destruction through a vascular address. Proc Natl AcadSci U S A 2002;99:1527–31.

[12] Bassett SE, Fennewald SM, King DJ, Li X, Herzog NK, Shope R, et al. Combi-natorial selection and edited combinatorial selection of phosphorothioateaptamers targeting human nuclear factor-kappaB RelA/p50 and RelA/RelA.Biochemistry 2004;43:9105–15.

[13] Gabizon A, Shmeeda H, Horowitz AT, Zalipsky S. Tumor cell targeting ofliposome-entrapped drugs with phospholipid-anchored folic acid–PEGconjugates. Adv Drug Deliv Rev 2004;56:1177–92.

[14] Steffan AM, Gendrault JL, McCuskey RS, McCuskey PA, Kirn A. Phagocytosis, anunrecognized property of murine endothelial liver cells. Hepatology1986;6:830–6.

[15] Langeggen H, Namork E, Johnson E, Hetland G. HUVEC take up opsonizedzymosan particles and secrete cytokines IL-6 and IL-8 in vitro. FEMS ImmunolMed Microbiol 2003;36:55–61.

[16] Dini L, Lentini A, Diez GD, Rocha M, Falasca L, Serafino L, et al. Phagocytosis ofapoptotic bodies by liver endothelial cells. J Cell Sci 1995;108:967–73.

[17] Kirsh T, Woywodt A, Beese M, Wyss K, Park J-K, Erdbruegger U, et al.Engulfment of apoptotic cells by microvascular endothelial cells inducesproinflammatory responses. Blood 2007;109:2854–62.

[18] Bengualid V, Hatcher VB, Diamond B, Blumberg EA, Lowy FD. Staphylococcusaureus infection of human endothelial cells potentiates Fc receptor expression.J Immunol 1990;145:4279–83.

[19] Canham LT. Bioactive silicon structure fabrication through nanoetching tech-niques. Adv Mater 1995;7:1033–7.

[20] Godin B, Gu J, Serda RE, Ferrati S, Liu X, Chiappini C, et al. Multistage meso-porous silicon-based nanocarriers: biocompatibility and controlled degrada-tion in physiological fluids. New York City, NY: 35th Annual Meeting andExposition of the Controlled Release Society; 2008. p. 575.

[21] Ferrari M. Nanogeometry beyond drug delivery. Nat Nanotechnol 2008;3:131–2.[22] Decuzzi P, Lee S, Bhushan B, Ferrari M. A theoretical model for the margination

of particles within blood vessels. Ann Biomed Eng 2005;33:179–90.[23] Decuzzi P, Lee S, Decuzzi P, Ferrari M. Adhesion of microfabricated particles on

vascular endothelium: a parametric analysis. Ann Biomed Eng 2004;32:793–802.[24] Sanhai WR, Sakamoto JH, Canady R, Ferrari M. Seven challenges for nano-

medicine. Nat Nanotechnol 2008;3:242–4.[25] Cohen MH, Melnik K, Boiarski AA, Ferrari M, Martin FJ. Microfabrication of

silicon-based nanoporous particulates for medical applications. BiomedMicrodevices 2003;5:253–9.

[26] Krysko DV, Brouckaert G, Kalai M, Vandenabeele P, D’Herde K. Mechanisms ofinternalization of apoptotic and necrotic L929 cells by a macrophage cell linestudied by electron microscopy. J Morphol 2003;258:336–45.

[27] Hartig SM, Greene RR, Carlesso G, Higginbotham JN, Khan WN, Prokop A, et al.Kinetic analysis of nanoparticulate polyelectrolyte complex interactions withendothelial cells. Biomaterials 2007;28:3843–55.

[28] Hartwig JH, Stossel TP. Interactions of actin, myosin, and an actin-bindingprotein of rabbit pulmonary macrophages. III. Effects of cytochalasin B. J CellBiol 1976;71:295–303.

[29] Stockem W, Hoffmann HU, Gruber B. Dynamics of the cytoskeleton in Amoebaproteus. I. Redistribution of microinjected fluorescein-labeled actin duringlocomotion, immobilization and phagocytosis. Cell Tissue Res 1983;232:79–96.

[30] Coleman DL. Regulation of macrophage phagocytosis. Eur J Clin Microbiol1986;5:1–5.

[31] Pan LF, Kreisle RA, Shi YD. Detection of Fcgamma receptors on human endothelialcells stimulated with cytokines tumour necrosis factor-alpha (TNF-alpha) andinterferon-gamma (IFN-gamma). Clin Exp Immunol 1998;112:533–8.

[32] Alberto MF, Bermejo EI, Lazzari MA. Receptor expression for IgG constant frac-tion in human umbilical vein endothelial cells. Thromb Res 2000;97:505–11.

[33] Vielma S, Virella G, Gorod A, Lopes-Virella M. Chlamydophila pneumoniaeinfection of human aortic endothelial cells induces the expression of FCgamma receptor II (FcgammaRII). Clin Immunol 2002;104:265–73.

[34] Arvidsson S, Askendal A, Tengvall P. Blood plasma contact activation onsilicon, titanium and aluminium. Biomaterials 2007;28:1346–54.

[35] Leroux JC, De JF, Anner B, Doelker E, Gurny R. An investigation on the role ofplasma and serum opsonins on the internalization of biodegradable poly(D,L-lactic acid) nanoparticles by human monocytes. Life Sci 1995;57:695–703.

[36] Klein CL, Bittinger F, Kohler H, Wagner M, Otto M, Hermanns I, et al.Comparative studies on vascular endothelium in vitro. 3. Effects of cytokineson the expression of E-selectin, ICAM-1 and VCAM-1 by cultured humanendothelial cells obtained from different passages. Pathobiology 1995;63:83–92.

[37] Scholz D, Devaux B, Hirche A, Potzsch B, Kropp B, Schaper W, et al. Expressionof adhesion molecules is specific and time-dependent in cytokine-stimulatedendothelial cells in culture. Cell Tissue Res 1996;284:415–23.

[38] Campbell RB, Fukumura D, Brown EB, Mazzola LM, Izumi Y, Jain RK, et al.Cationic charge determines the distribution of liposomes between thevascular and extravascular compartments of tumors. Cancer Res2002;62:6831–6.

[39] Borchard G, Kreuter J. The role of serum complement on the organ distribu-tion of intravenously administered poly(methyl methacrylate) nanoparticles:effects of pre-coating with plasma and with serum complement. Pharm Res1996;13:1055–8.

[40] Sakhalkar HS, Dalal MK, Salem AK, Ansari R, Fu J, Kiani MD, et al. Leukocyte-inspired biodegradable particles that selectively and avidly adhere to inflamedendothelium in vitro and in vivo. Proc Natl Acad Sci U S A 2003;100:15895–900.

[41] Muller WA, Weigl SA, Deng X, Phillips DM. PECAM-1 is required for trans-endothelial migration of leukocytes. J Exp Med 1993;178:449–60.

[42] Schenkel AR, Mamdouh Z, Chen X, Liebman RM, Muller WA. CD99 playsa major role in migration of monocytes through endothelial junctions. NatImmunol 2002;3:116–8.

R.E. Serda et al. / Biomaterials 30 (2009) 2440–24482448