ARTICLE IN PRESS

0142-9612/$ - se

doi:10.1016/j.bi

�Correspond271-3181.

E-mail addr

Biomaterials 26 (2005) 2061–2072

www.elsevier.com/locate/biomaterials

Epithelial internalization of superparamagnetic nanoparticles andresponse to external magnetic field

Kenneth Dormera,�, Charles Seeneyb, Kevin Lewellingc, Guoda Liand, Donald Gibsonc,Matthew Johnsond

aDepartment of Physiology, Oklahoma University Health Sciences Center, 940 S.L. Young Blvd., Room 634, Oklahoma City, OK 73190, USAbChemistry Department, NanoBioMagnetics Inc., 124 N. Bryant Ave., Suite C3, Edmond, OK 73034, USAcPhysics Department, Oklahoma Christian University, 2501 E. Memorial Rd, Edmond, OK 73013, USA

dDepartment of Physics and Astronomy, University of Oklahoma, 131 Nielsen Hall, Norman, OK 73019, USA

Received 5 April 2004; accepted 25 June 2004

Available online 10 August 2004

Abstract

Superparamagnetic magnetite nanoparticles (MNP) coated with silica were synthesized and chronically implanted into the middle

ear epithelial tissues of a guinea pig model (n ¼ 16) for the generation of force by an external magnetic field. In vivo limitations of

biocompatibility include particle morphology, size distribution, composition and mode of internalization. Synthesis of MNP was

performed using a modified precipitation technique and they were characterized by transmission electron microscopy, X-ray

diffractometry and energy dispersive spectroscopy, which verified size distribution, composition and silica encapsulation. The

mechanism for internalizing 1672.3 nm diameter MNP was likely endocytosis, enhanced by magnetically force. Using sterile

technique, middle ear epithelia of tympanic membrane or ossicles was exposed and a suspension of particles with fluoroscein

isothiocyanate (FITC) label applied to the surface. A rare earth, NdFeBo magnet (0.35T) placed under the animal, was used to pull

the MNP into the tissue. After 8 days, following euthanasia, tissues were harvested and confocal scanning laser interferometry was

used to verify intracellular MNP. Displacements of the osscicular chain in response to an external sinusoidal electromagnetic field

were also measured using laser Doppler interferometry. We showed for the first time a physiologically relevant, biomechanical

function, produced by MNP responding to a magnetic field.

r 2004 Elsevier Ltd. All rights reserved.

Keywords: Nanoparticles; Middle ear; Biomechanics; Hearing; Endocytosis

1. Introduction

Superparamagnetic nanoparticles (MNP), such asmagnetite, have been widely used for biomedicalapplications [1]. Utilization of MNP to produce forcesin living cells, likewise, is not entirely a new concept [2].Magnetic twisting cytometry was previously developedto generate torque on cells in culture to assess the role ofmechanical stress during development, notably indeveloping pulmonary epithelium [3–5]. Biomedical

e front matter r 2004 Elsevier Ltd. All rights reserved.

omaterials.2004.06.040

ing author. Tel: +1-405-271-2226-1221; fax: +1-405-

ess: ken-dormer@ouhsc.edu (D. Kenneth).

applications of MNP include magnetic resonanceimaging contrast enhancement, hyperthermia, intravas-cular targeted delivery of therapeutics, biosensors andothers [1,6–9]. Targeted delivery of therapeutics requiresadequate particle susceptibility to and directionalcontrol by an external magnetic field. Chronicallyimplanted MNP for assisted biomechanical organ ortissue functions is feasible only if particle susceptibility iscommensurate with directionality and strength ofexternal magnetic fields [2]. Physiological mechanismsfor cellular internalization of particles include fluidphase or receptor-mediated endocytosis, phagocytosisand non-endocytic pathways. Enhancement of MNPinternalization by magnetic forces and subsequent long-

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–20722062

term viability are not understood [1]. Additionally, thelimits of intracellular MNP mass sufficient to exertforces without harming cells are unknown. Little is alsoknown of intracellular trafficking following magneti-cally enhanced endocytosis and potential iron cytotoxi-city from particle degradation [10]. Additionally, theeffects of cytosolic MNP on cellular apoptotic processesare untested.In recent years electromagnetic implantable hearing

devices for patients with sensorineural hearing loss havebeen developed. Permanent magnets, such as neody-mium–iron–boron, are surgically implanted into themiddle ear, wherein the implants interact with externalelectromagnetic fields mechanically directly driving theossicular chain [11]. Resulting vibrations representingacoustic input are conveyed into the inner ear (cochlea),creating the percept of sound [12]. Such ‘‘direct drive’’hearing device technology has generated high fidelity,amplitude and frequency sounds. Since implanted hardmagnets can generate sufficient force to produce sound,we hypothesized that comparable vibratory forces couldbe produced by MNP, with sufficient mass implanted inthe ossicular epithelium or tympanic membrane. Per-haps, hearing amplification could serve as a model oftissue biomechanics from implanted MNP respondingto an external magnetic field.Of prospective mechanisms for internalization, we

hypothesized that endocytic processes will be enhancedby forces pulling silica encapsulated MNP into thetympanic membrane and middle ear epithelium. Widderand colleagues have shown in magnetic drug targetingstudies that external magnetic fields can concentrateMNP in target tissue, enhancing cell entry by endocy-tosis [13]. By contrast, intravascular MNP targeting ofendothelial receptors must contend with opsonizationby proteins and phagocytosis by the reticuloendothelialsystem. Hence, coatings such as polyethylene glycol(PEG) have been used to reduce macrophage recogni-tion and uptake [1]. Silica encapsulation of MNP withnegative surface charge (zeta potential) is useful forcorrosion protection, colloidal dispersion and as sub-strate for particle functionalisation [14,15]. Silica doesnot substantially interfere with magnetic susceptibility[16,17]. We directed forces on silica encapsulated MNPusing an external magnetic field to more effectivelyinternalize these particles, irrespective of endocytic andnon-specific mechanisms [1,18].A requirement for MNP generating forces in cells and

tissues is long-term viability. Unlike MNP for drugdelivery, chronic implantation requires cell compatibility,hermetic encapsulation, and no exocytosis. Potential ironleaching and toxicity, an aspect of long-term MNPviability, has limited data. Normal intracellular ironhomeostasis involves internalization of the complex ofironbound transferrin via the transferrin receptor butexcessive intracellular iron ions are potentially toxic

through the formation of oxygen radicals and peroxidativedamage [10,19]. Intracellular iron balance could be upset ifMNP were to leach iron. Nevertheless, studies on magneticresonance imaging showed (I.V.) ferrite nanoparticles ascompatible with hepatic reticuloendothelial cells andcausing no hepatocellular injury [20].We report here on a model for generation of forces in

living tissues, implantation of superparamagnetic nano-particles in the middle ear epithelium. Magnetitenanoparticles (Fe3O4) were synthesized by a modifiedMassart technique [21] and silica encapsulated forhermeticity and functionalisation. Particles were char-acterized using X-ray diffraction (XRD) and transmis-sion electron microscopy (TEM) equipped with energydispersive X-ray spectroscopy (EDS). Middle earepithelial cell internalization was enhanced by anexternal magnetic field and confirmed by observingconjugated fluoroscein isothiocyanate (FITC) intracel-lular fluorescence under confocal microscopy. Neitherthe mechanisms of MNP endocytosis nor their intracel-lular trafficking were the focus of this study.The ability to generate sinusoidal, vibratory forces by

the interaction of MNP with an external magnetic fieldwas also tested. The tip of an electromagnetic coil wasplaced 1–2mm from either the implanted tympanicmembrane or incus epithelium. One and two kHzsinusoidal vibration of the ossicular chain was pro-duced. Thus, we concluded that intracellular MNP,implanted up to 15 days, remained viable in situ andcould be used to produce ossicular vibrations consistentwith the perception of sound in humans.

2. Materials and methods

2.1. Synthesis and characterization of magnetite

nanoparticles (MNP)

2.1.1. Preparation of magnetic nanoparticles

Nanoparticle synthesis employed criteria for size,superparamagnetism, mass, hermetic encapsulation,substrate for linkers and viability in tissues. Minimalsize optimizes ease of entry across cell membranes[21,22]. To enhance endocytosis by magnetic forces,MNP less than 30 nm diameter were sought. Magnetiteless than 30–50 nm also exhibits superparamagnetismdue to single domain crystalline structure. Superpar-amagnetism exhibits no remanence, dispromoting ag-glomeration that occurs with magnetized particles Thespherical magnetite particles (Fe3O4) with relatively highmagnetic susceptibility included an �5 nm shell of silica(SiO2). Encapsulation of the magnetite by SiO2 prohib-ited corrosion and provided an anionic surface chargethat promoted endocytosis [15] as well as a substrate forattachment of amines that can serve as linkers to othermolecules.

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–2072 2063

To synthesize MNP we used a modified procedure ofMassart [21,23] in an oxygen-free medium to preventformation of maghemite. A 2M iron (II) sulfateheptahydrate solution was prepared in 2M HCl andcombined with 1M iron (III) chloride hexahydrateaqueous solution in a molar ratio of 2:1 Fe3+:Fe2+.The solution was mixed and added to 0.7M ammoniumhydroxide with rapid stirring. The resulting precipitatewas stirred for 30min, restrained using a permanentmagnet (300 gauss) and the supernatant was discarded.After several washes, the precipitate was re-suspended in0.7M ammonium hydroxide and peptized through theaddition of 1M tetramethylammonium hydroxide ali-quots. The total volume was then taken to 250ml.

2.1.2. Surface modification of magnetite nanoparticles

with silica

The particles were next coated with SiO2 [24]. Asuspension of magnetite nanoparticles was vigorouslystirred and a 4ml aliquot taken to a volume of 100mlwith distilled water. A solution of 0.54% sodium silicatewas prepared at pH 10.5, and 4ml was added to thepreviously prepared MNP suspension. The suspensionwas pH adjusted to 10.0 then stirred for 2 h and allowedto stand for 4 days. Excess silica was removed by severalwashes, a magnet retaining the MNP. To funtionalisethe silica coating with amine groups, particles were nexttreated with 3-aminopropyltrimethoxysilane. A 1mlaliquot of the suspension was brought to a volume of5ml with distilled water and sufficient 3-aminopropyl-trimethoxysilane added for a final concentration of 5%[25]. The reaction system was stirred at room tempera-ture for an hour. After the incubation period, theparticles were again washed with distilled water.The Kaiser assay [26] was used to confirm the

presence of free amines on the silica-encapsulated,amine-functionalised MNP. A sample of particles wasreacted with 0.28 M ninhydrin, 76% phenol in ethanoland 0.0002M KCN. After a short incubation in boilingwater, a blue color indicates the presence of free amines(yellow color indicates the absence of amines).

2.1.3. Functionalisation of silica coated magnetic

nanoparticles

Using the amine linker, FITC was conjugated to theMNP for subsequent location in cells using confocalfluorescence microscopy. Particles were functionalisedfollowing standard conjugation protocol (MolecularProbes, 2003, Eugene, OR). After the conjugation,several rinses with sterile 0.9% saline removed excessnon-conjugated, FITC and promoted sterility. Theparticles were stored in saline at 20 1C until surgery.

2.1.4. Nanoparticle characterization

2.1.4.1. XRD analysis. The MNP crystal structure,morphology, size distribution, elemental and chemical

composition were investigated using XRD, TEM, andEDS. Crystal structure (phase) and average nanoparticlesize found in a macroscopic sample can be determinedusing XRD techniques. XRD analyses were performedusing a diffractometer (Scintag X’TRA, Applied Re-search laboratories, Ecublens, Switzerland).

2.1.4.2. TEM and EDS analysis. For this work,crystallinity, phase and crystal size are important proper-ties because they affect particle magnetic susceptibility.TEM analysis allows individual particles to be directlyimaged, whereupon images are used to determine thespecific morphology, chemical composition, as well asparticle size and size distribution. High-resolution TEM(HRTEM) and electron diffraction can show whether anindividual MNP is a single crystal or polycrystalline.Under the appropriate imaging conditions, silica-coatediron-oxide nanoparticles can be differentiated fromuncoated particles. The elemental constituents can alsobe unambiguously determined using EDS. The TEMexperiments, including selected area electron diffraction(SAED), HRTEM and EDS, were all performed using aJEOL 2000FX electron microscope (JEOL, Japan)operated at 200 kV equipped with an EDS system(KEVEX, Thermo Kevex X-ray, Scotts Valley, CA).Initial assessment of the nature and extent of silica

surface treatments was tested. Particle samples thatranged in silica content, based on reaction feed ratioand time conditions, were dispersed in 0.9% saline andthe settling rates were observed as a function of time.Particles with higher ratios in the reaction feed have thelongest settling rates. Hermeticity of the silica coating andnascent resistance to corrosion of uncoated magnetitenanoparticles was tested by soaking both uncoated andsilica-coated nanoparticles in 10% NaCl solution for 2days at room temperature and then 2 days at 40 1C (totalof 4 days). A standard Sodium Thiocyanate (NaSCN)assay was used periodically over 96h to reveal thepresence of Fe ions. Up to 6 months later tests forcorrosion were also performed on the MNP stocksolution, an unprecipitated ferrofluid in tetramethylam-monium hydroxide. Additionally, precipitated MNP thatwere placed in aqueous solution, double distilled water atpH=13 were tested after 5 months of soaking using theNaSCN test for the Prussian Blue Reaction.Magnetic susceptibility (emu/g) of both unmodified

and silica modified MNP were measured using VibratingSample Magnetometry (VSM, ADE/DMS Model 880,Arkival Technologies, Nashua, NH) which provides aquantitative measure and is necessary for assessing theability of MNP to respond to an external magneticfield [27].The zeta potential for the SiO2 encapsulated MNP,

without the amine linkers used to conjugate FITC, wasmeasured by dispersing the particles in distilled water atpH=7.5 adjusted using sodium hydroxide. The MNP

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–20722064

suspension was sonicated in a water bath for 15min thendiluted into concentrations of 42, 40 and 38mg/ml,sonicated again and their electrophoretic mobility mea-sured to determine zeta potentials (Zeta-Meter Inc.,Stanton, VA).

2.2. Cellular uptake of magnetite nanoparticles

2.2.1. Surgical approach

Guinea pigs (n ¼ 16) of either sex, weighing 375–450 gwere anesthetized with Ketamine and Xylazine (35 and15mg/kg IP, respectively). Skin surrounding the left orright pinna was shaved and sterilized and using steriletechnique, a retro-auricular incision exposed the tem-poral bone over the middle ear cavity. The middle earspace was drilled opened, exposing the incus as thetarget for implantation (MicroCraftTM surgical drill,Xomed Inc., Jacksonville, FL).A sterile solution of MNP in physiological saline at

pH=7.4 was sonicated (Sonicor, Copaigue, NY) for3–4min before implantation. Next, 25 ml of the MNPsolution, adherent to the tip of a gentle curved pick, wastouched to the tissue surface. This application wasrepeated to place a total volume of 50–75 ml on either theincus or tympanic membrane. Nine animals wereimplanted in the incus epithelium (n ¼ 3 controls,contralateral ear). Two animals were implanted on thetympanic membrane (n ¼ 1 control, contralateral ear).The operative skin site was closed using 3-0 absorbablesuture and the animals were monitored during simulta-neous external magnetic field exposure and recoveryfrom anesthesia.

2.2.2. Epithelial internalization of nanoparticles

During recovery from anesthesia the animal’s headwas laid on the pole face of a 4 in cube permanentmagnet (NdFeBo, 50 MGO) with the implanted earfacing upward. Thus, the MNP solution held by surfacetension was magnetically pulled downward, into theepithelia. The MNP solution on epithelia at distances of1 in from the pole face of the magnet experienced a fieldof �0.35T. Each animal was exposed to this magneticfield for 20–30min then was subsequently returned to itscage for 1–15 days of monitored recovery.

2.3. Histology and fluorescence microscopy

Recovering 8–15 days, animals were re-anesthetizedand euthanasia caused by an injection of sodiumpentobarbital (6 grains, IM). The experimental incuswas removed and placed in 80% ethanol, 10% normalsaline for a minimum of 24 h. Next, the incii weredecalcified (Decalcification SolutionTM, Richard AllenScientific, Kalamazoo, MI) for 3 days then processed forparaffin sectioning (Tissue-Tek VIPTM, Sakura Finetek,Torrance, CA). Aldehydes were excluded to prevent

autofluorescence of tissues. Transverse 5 mm microtomesections of the incii or tympanic membrane in paraffinblocks were mounted and either stained with hematox-ylin and eosin for histopathology or unstained forconfocal laser microscopy.Transverse epithelial sections of a decalcified incus

from experimental ears having been implanted for 5days were given to a Veterinary Pathologist whocompared the (unlabeled) experimental and controltissues using light microscopy (40–100� ). Unstainedtransverse sections also were examined using an Argonlaser scanning Spectral Confocal and MultiphotonMicroscope (Leica Model TCS SP2, Leica Microsys-tems, Mannheim, Germany). Stacks of 10–20 scans weremade through the epithelium lining the incii, or thetympanic membrane epithelium while looking forintracellular fluorescence, indicating the presence ofMNP.

2.4. Laser Doppler interferometry measurement of

middle ear displacements

Confirmation of epithelial implantation with MNPwas made using laser Doppler interferometry (LDI),providing contactless velocity and displacement mea-sures with a frequency range of 0–150 kHz [28].Following euthanasia, the animal was placed in a lateralrecumbent position, implanted ear up. The pinna wasremoved for visualization of the bony ear canal andtympanic membrane through an operating microscope.The surgical site was reopened and a 1� 1mm2

reflective tape (3-M, Minneapolis, MN) placed eithernear the umbo on the tympanic membrane or on thelateral incus. Middle ear displacements were measuredusing single point, helium neon laser LDI (Model OFV501 and Model 3000 Controller, Polytec PI, Tustin,CA). When an electromagnetic coil (8mm length, 2mmwidth, 6.5mH) was placed 1–2mm from the epithelialsurface and activated with 500 or 1000Hz sine waves at5–8V, peak to peak (Model 80 Function Generator,Wavetek, San Diego, CA and Model 2706 PrecisionAmplifier, Bruel & Kjaer, Denmark), the resultingmagnetic field vibrated the MNP. The middle ear wasintact in these studies; therefore, LDI displacementmeasurements reflected movements of the ossicularchain.

3. Results

3.1. MNP characterization

In order to confirm the iron oxide phase and size ofthe core in the MNP, uncoated, bare magnetite particleswere characterized before applying a SiO2 shell. Fig. 1shows an XRD spectrum of the uncoated MNP. The

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–2072 2065

spectrum displays peak positions (labeled) and relativeintensities consistent with the crystal structure ofmagnetite, Fe3O4. The broad background peak is fromthe glass substrate used during XRD scanning. XRDanalyses confirmed the absence of other common phasesof iron oxide species. XRD analysis determined thediameter of these uncoated MNP, using Scherrer’sformula [29], with an average of 10 nm. Fig. 2 showsTEM results for uncoated MNP with a typical diameterof 10 nm, in agreement with the XRD results. Selectedarea electron diffraction (SAED) patterns in Fig. 2bconfirm the phase of the uncoated MNP to be crystallinemagnetite, again in agreement with the XRD analysis.

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75

(220)

(311)

(400)

(511)(440)

2θ (o)

Inte

nsit

y (a

.u.)

Fig. 1. XRD pattern of uncoated superparamagnetic, ferrite nano-

particles; the indexing corresponds to Fe3O4 magnetite.

Fig. 2. (a) Transmission electron micrograph of uncoated magnetite nanop

showing high-resolution electron micrograph of an individual nanoparticles

Finally, Fig. 2c is an HRTEM image of an individualparticle as marked in Fig. 2a. This image clearly showslattice planes with spacing of 4.8 A, corresponding toFe3O4 magnetite (111 lattice spacing).Silica-coated magnetite particles were also investi-

gated using TEM and EDS. Fig. 3a is a TEMmicrograph of an ultra-thin (o15 nm) layer of silica-coated MNP. For unambiguous statistical analysis ofMNP it is crucial to prepare such ultra-thin layers, sothat individual rather than aggregates of nanoparticlesare imaged. For these silica-coated particles we reducedagglomeration by using a polyelectrolyte, polydiallyldi-methylamonium chloride (PDDA), to positively chargethe ultra-thin carbon coated, copper TEM grid surfacebefore applying a dilute, basic, aqueous suspension ofnegatively charged MNP. Statistical analysis of coatedNPs indicated an average particle diameter of 16 nmwith a standard deviation of 2.3 nm (n ¼ 80). Fig. 3c isan enlarged view of a sampled individual particle, whichshows a magnetite core (dark) covered by a layer ofsilica (light). Figs. 4a and b show the X-ray spectrataken from shell and core regions of a silica-coatedparticle as determined from a HRTEM image. Thespectrum from the silica shell shows very low ironcontent, whereas that from the core shows much higheriron content. The small amount of iron seen in the SiO2coating is not likely from iron present in the SiO2 shell,but rather from nearby iron oxide excited due toelectron beam-broadening inherent in the EDS techni-que. Based on the EDS analysis, it was confirmed thatthe outer shell is SiO2 and the core is iron oxide asexpected.Zeta potentials (n ¼ 6) of silica-coated MNP ranged

from �15 to �20mV, though 30mV is ideal according

articles. (b) SAED pattern from large area of NPs. (c) Magnification

as marked in (a).

ARTICLE IN PRESS

Fig. 3. (a) Transmission electron micrograph of silica-coated superparamagnetic, magnetite nanoparticles; size distribution of coated nanoparticles is

shown in (b) and detailed core-shell structure of an individual MNP is shown in (c).

Energy (keV)

Cou

nts

(a.u

.)

EDS on Dark Core of Particle

0

50

100

150

200

250

0 1 2 3 4 5 6 7 8 9 10

Fe Kα

Fe

Kβ

Cu Kα

Cu

Kβ

SiK

O K

Fe

L

C K

(b)

EDS on Silica Shell of Particle

Energy (keV)

Cou

nts

(a.u

.)

(a)

SiK

Fe

Kα

O K

Cu Kα

Cu

Kβ

0

50

100

150

200

250

0 1 2 3 4 5 6 7 8 9 10

Cu

L

Fig. 4. Energy dispersive X-ray spectroscopy spectra from a silica shell

(a) and the iron oxide core of silica-coated nanoparticles (b).

Kenneth et al. / Biomaterials 26 (2005) 2061–20722066

to Widder and colleagues [13]. Nevertheless, theparticles remained dispersed in aqueous solution for1 h and did not show visible precipitation until 24 hlater, indicating electrostatic repulsive charge moder-ately cancelled van der Walls attraction forces. Magneticsusceptibility of the MNP used in this study had valuesof 31 and 23 emu/g for unencapsulated and silica-encapsulated MNP, respectively.

Saline corrosion testing of the MNP by the NaSCNPrussian Blue Reaction showed that 48 h exposure tocorrosive saline caused no leaching of iron or particledegradation. Fig. 5a shows control NaSCN test resultfollowing 4 days exposure of commercially obtainedferrite nanopowder in 10% saline (Tal Materials, AnnArbor, MI). After 4 days the presence of Fe ions in themedia was revealed by the NaSCN reaction. Fig. 5bshows the negative Prussian Blue Reaction, testedconcurrently on unencapsulated MNP of the type usedin this study with the addition of silica encapsulation.No Fe ion leaching occurred, indicating the magnetitecrystalline structure was nonreactive and suggesting thatsilica coating would provide additional protectionagainst corrosion during short-term implantation. How-ever, both the silica-coated MNP in the ferrofluid stateand those in pH adjusted double distilled water didshow evidence of long-term corrosion, following 6 and 5months of soaking, respectively. Thus, long-termhermeticity of these MNP must be demonstrated forlong-term viability in vivo.

3.2. Uptake of MNP by epithelial cells

Using confocal laser and epifluorescence microscopywe showed that FITC-labeled MNP were present inepithelial cells of the incus and tympanic membraneafter 20min of magnetic field exposure. Quantificationof label or number of particles was not performed butincreased fluorescence on the upper surface of the inciiindicated a moderate mass of MNP had been inter-nalized (Fig. 6a) as well as in the tympanic membrane(Fig. 6b). Epithelia from control ears, not exposed to theexternal magnetic gradients, showed reduced intracel-lular fluorescence, suggesting that endocytosis is also

ARTICLE IN PRESS

Fig. 5. The Prussian Blue Reaction was initiated to test hermeticity of the coated magnetite. In (a) bare ferrite nanoparticles elicited the reaction

while in (b) the test was negative on the silica-coated magnetite particles, demonstrating no free iron and corrosion protection.

Kenneth et al. / Biomaterials 26 (2005) 2061–2072 2067

internalizing MNP. Fluorescence labeling 15 daysfollowing implantation suggested that cells had nottotally expelled the MNP by exocytosis.Silica encapsulated magnetite in the cytosol was

apparently non-toxic as no indications of cell pathologywas noted in implanted incii epithelia relative to controltissues. No giant cells or other indicators of inflamma-tory processes were noted no visible indications ofapoptosis or necrosis were seen following 2–15 days ofimplantation. A freshly explanted incus from a guineapig following 8 days of implantation with MNP isshown in Fig. 7. The dark coloration is from implantedMNP on the lateral (surgically upward) surface of theapparently healthy incus.

3.3. Vibration of the ossicular chain

Implanted epithelia of incus and tympanic membraneresponded to an external magnetic field and could bemeasurably vibrated at 2000Hz. When exposed to theelectromagnetic sine wave (1000Hz), frequency dou-bling occurred in the ossicular vibration due to thesuperparamagnetic property of MNP. Particles at-tracted to both north and south polarities produced asine wave vibration of 2000Hz. The recorded frequencywas a pure doubling of the sinusoidal input, with nodiscernable distortion. Table 1 shows the coil activationparameters and resulting laser interferometer displace-ments as measured from the tympanic membrane orincus in three guinea pigs implanted with MNP.

4. Discussion

4.1. Characterization of MNP for generation of forces in

tissues

XRD and TEM analyses indicated that uncoatedMNP had a uniform size distribution and magnetitephase. The average size of unencapsulated MNP wasabout 10 nm and HRTEM images indicated that theywere single crystal domain. For the silica-coated MNP,TEM images and EDS spectra showed most MNP werecoated by amorphous silica with an average size of1672.3 nm. This size is below the 20–30 nm criteria forachieving optimal membrane binding [30,31].Agglomeration impairs the implantation of nanopar-

ticles using external magnetic forces, as aggregates ofparticles are less likely to cross cell membranes.Magnetite nanoparticles naturally agglomerate in aqu-eous solutions at physiological pH levels. Greater netsurface charge lessens agglomeration. The isoelectricpoint for bare magnetite nanoparticles, as measured bythe Zeta-Meter, is 6.8 within the pH range of 6–10.Previous studies have shown that the isoelectric point isshifted by silica encapsulation, increasing the zetapotential and decreasing the propensity to agglomerate[14,32]. Absolute zeta potentials 430mV generallyassure mutual repulsion of MNP and no agglomeration[13,33]. Our MNP placed on the middle ear epitheliumhad sufficient surface charge to create a nominallyhomogenous dispersion at pH44, which contributed tothe effective internalization in the guinea pig model.

ARTICLE IN PRESS

Fig. 6. (a) A 40� confocal laser micrograph is shown from the single cell epithelial layer of the incus epithelium with fluorescent magnetite

nanoparticles; the ossicle was decalcified and the soft tissue then processed and sectioned to reveal fluorescenct intracellular nanoparticles. The light

areas of the tissue depict intracellular aggregates of nanoparticles. (b) A 100� confocal scanning laser micrograph of a section from guinea pig

tympanic membrane having been implanted for 8 days with FITC-labeled MNP. The light areas depict intracellular locations of aggregates of

nanoparticles.

Kenneth et al. / Biomaterials 26 (2005) 2061–20722068

Dispersion was important, hence, for implantation allparticles were also sonicated in isotonic saline atpH=7.4.Key criteria for chronic implantation of MNP and

production of force involved the ferrite formula, its massand magnetic susceptibility. Minimizing size is impor-tant for ease of cellular internalization, but reducessusceptibility [34]. Theoretically, magnetite crystals canhave susceptibilities as high as 90 emu/g, depending onsuch factors as synthesis, isolation conditions and

particle size. In biological applications, other investiga-tors have used polymer-coated magnetic nanoparticlesin the range of 30 emu/g, which is comparable to ourmodel [27]. Our somewhat lower susceptibilities wereprimarily are due to the silica encapsulation.Long-term hermeticity provided by SiO2 encapsula-

tion remains unclear. Related in vitro testing showedbare magnetite nanoparticles did not corrode whilesuspended in 10% saline over 4 days. Such corrosionresistance was likely due to the tight crystalline structure

ARTICLE IN PRESS

Fig. 7. Incus of guinea pig explanted following 8 days of implantation. Darker lateral surface indicated by the arrow shows where magnetite

nanoparticles were placed intraoperatively and then subsequently internalized by an external magnetic field. Implanted epithelia did not display any

inflammatory responses when examined by a histopathologist.

Table 1

Laser Doppler interferometry data on displacements of middle ear epithelia containing superparamagnetic nanoparticles

Animal Epithelium Excitation freq. (Hz) Frequency out (Hz) Voltage to coil (V) Displacement P�P� 10�4m Displacement (A)

GP17 Incus 500 1000 5.5 0.106 0.12

1000 2000 0.028 0.03

GP18 Incus 500 1000 5.5 0.106 0.12

1000 2000 0.067 0.07

GP24 Tympanic membrane 1000 2000 8.0 0.165 0.17

In three animals the ability to vibrate implanted middle ear tissues by an external magnetic field was verified using single point interferometry. Note

that the tympanic membrane was vibrated more easily than the incii, consistent with its greater compliance as a membrane unattached to a substrate.

Vibrations were obtained up to 3200Hz.

Kenneth et al. / Biomaterials 26 (2005) 2061–2072 2069

of magnetite. Nevertheless, after 5–6 months in salineboth naked and silica-encapsulated particles showedsome corrosion (detection of free iron). Our biomecha-nical assay, ossicular chain vibrations in vitro by anexternal magnetic field, was effective after 8 days ofimplantation. One explanation for our successful resultsis that the majority of the MNP were hermetically sealedand other particles, incompletely encapsulated, were notimplanted long enough for corrosion to occur. Certainlyfor chronic implantation, 100% hermeticity of MNPwould be required.

4.2. Internalization of MNP by magnetic forces

Physiological intracellular trafficking of substancesbegins with internalization and culminates with exocy-tosis. If MNP were internalized solely by endocytosis,then lysosomal pH of 5–5.5 could potentially promotecorrosion of SiO2 encapsulation [14,17]. Endocytic

processes normally are concentration-, time- and en-ergy-dependent with internalization occurring afterabout 1min of incubation in cell culture. Other studieshave shown that nanoparticle exocytosis begins once anextracellular concentration gradient is removed, withabout 65% of an internalized fraction of particles(9773 nm diameter) undergoing exocytosis in 30min[35]. In our studies, silica encapsulation (negative zetapotential) should repulse negative membrane surfacecharges. Our MNP aminated (positive surface charge)for FITC conjugation should be attractive to membranesurfaces, just as fusegenic cationic polymers promoteinternalization [22]. A new class of albumin coated,anionic, maghemite nanoparticles, however, has shownhigh affinity for cell membranes with endocyticefficiency three times greater than dextran coatednanoparticles [15]. Other studies also have shownnegative nanoparticles non-specifically taken up morereadily than neutral or positive charged particles [36].

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–20722070

Irrespective of natural endocytosis and these conflictingstudies on non-specific, surface charge related interac-tions, internalization was greatly enhanced by magneticforces in these in vivo studies.Since nanoparticle retention increases with concentra-

tion in cell culture [35], it may be that our middle earepithelia were continually endocytic due to excess MNPadsorbed to the surfaces. Since MNP remained intra-cellular up to 15 days following implantation, magne-tically enhanced endocytic mechanisms may havecontributed to their substantial internalization. Whatremains to be proven is long-term viability of theseMNP, with sinusoidal generation of forces, and withinnormal apoptotic lifetimes of these tissues.

4.3. Generation of forces in tissues

For the first time we have shown that implantedsuperparamagnetic nanoparticles have produced sinu-soidal forces in vivo. Displacements of the guinea pigossicular chain at an auditory frequency were relativelygreater than those recorded from human ears. Inprevious studies using fresh-frozen temporal bones(n ¼ 17), 90 dB SPL calibrated sinusoidal 1000Hzsounds were presented 2mm from the surface oftympanic membranes and LDI measured averagedisplacements of 0.034 mm (peak to peak) for 1000Hzinput [28,37]. Incus and tympanic membrane displace-ments in this study using the apparatus were 0.106 and0.165 mm, respectively (Table 1). Though indicatingsubstantial amplification relative to the human, theguinea pig middle ear has much less mass and the coilvoltages used were greater than those typically used inhuman implantable hearing devices [11]. Nevertheless,hearing amplification was not the goal of this study, butrather feasibility of particle generation of force. Ourresults are consistent with biomechanics of the middleear in that greater displacements were recorded from thetympanic membrane (Table 1), which is consistent withthe ossicular lever ratio.Other applied forces using magnetic particle metho-

dology has been magnetic twisting cytometry, probingmicromechanical properties of epithelia in cell lines [38].Magnetic microbeads (5 mm) bound to cell surfaceligands applied sinusoidal (0.03–16Hz) torque in re-sponse to an external magnetic field [39]. Cell deform-ability is important to understanding cell motility,apoptosis or DNA synthesis, and in our study deform-ability of the middle ear epithelium relates to the long-term ability to generate acoustically relevant displace-ments.Long-term viability of MNP in middle ear epithelial

cells also depends upon the response of these cells tonewly imposed forces. Cells may use tensegrity archi-tecture to structure and stabilize themselves throughcontinuous tension distributed across the structural

elements [40]. Tensegrity cell models predict complexcell behaviors, including how cells change shape whenthey adhere to rigid extracellular matrices, as in theepithelium covering middle ear ossicles. Such modelsalso predict that cells and nuclei will immediatelyrespond to mechanical stresses transmitted over cellsurface receptors that physically connect the cytoskele-ton with the extracellular matrix [41]. Thus, transegritymodels would predict that chronic stresses imposed onthe middle ear epithelium will result in cell stiffnesschanges to accompany the continual internal stress onthe cytoskeleton during ossicular displacments.

4.4. Non-toxicity of MNP

Short-term viability of intracellular MNP was demon-strated by the presence of particles 15 days afterimplantation by the absence of histopathology and byeffective mechanical transduction of the middle ear.Magnetite microparticles have been shown to be lesssensitive to oxidation than magnetic transition metalparticles [27]. Nevertheless, long-term cell viability frompotential iron toxicity remains a concern, if silicaencapsulation should fail. Normal homeostasis of ironis primarily regulated at the level of messenger RNAtranslation, where regulatory proteins control uptake,storage and utilization [19]. Regarding longer in vitrostability, we noted oxidation of MNP only after 5months of soaking in hypertonic saline.The only data on in vivo iron particles showed non-

toxicity when ferrite nanoparticles were administered(I.V.) to patients for enhanced magnetic resonanceimaging. Iron determinations in the liver, spleen, lungand kidney after a massive 250mg iron/kg body weightdose showed that particles were taken up by hepaticreticuloendothelial cells, with no evidence of mitochon-drial or microsomal lipid peroxidation or organelledysfunction, sensitive indicators of iron-induced hepa-tocellular injury [20]. Incus MNP released in humans byexocytosis or apoptosis would pass down the Eustachiantube to the throat, be swallowed and eliminated fromthe body.

5. Conclusion

In summary, we have shown that silica encapsulated,magnetite nanoparticles can be used in conjunction withan external magnetic field to produce biomechanical forcesin the middle ear. MNP, customized for ease ofinternalization, and magnetic susceptibility can be im-planted in epithelia without producing toxic effects andremain viable for at least 15 days. Displacements in tactossicular chain at auditory frequencies have been gener-ated in the guinea pig using a miniature electromagneticcoil. Amplitudes of these displacements are comparable to

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–2072 2071

those produced by 90dB SPL at the tympanic membranein fresh frozen human temporal bones.

Acknowledgements

Our appreciation extends to Arif Mamedov, Ph.D. forassistance in synthesis, Wanda Day, HTL II for herhistological processing of tissues, Fadee Mondalek,B.S., for technical support, Don Nakmali, BSEE,Hough Ear Institute for assistance with interferometryand Stanley Kosanke, DVM, Ph.D., for evaluating thehistopathology. This work was performed under NIHSBIR Grant 1 R43 DC05528-01 to NanoBioMagneticsInc.Conflict of Interest statement:Charles E. Seeney and Kenneth J. Dormer are

Officers of NanoBioMagnetics Inc., recipient of theabove NIH-SBIR grant.

References

[1] Zhang Y, Kohle N, Zhang M. Surface modification of super-

paramagnetic magnetite nanoparticles and their intracellular

uptake. Biomaterials 2002;23:1553–61.

[2] Valberg PA, Butler JP. Magnetic particle motions within living

cells. Physical theory and techniques. Biophys J 1987;52:537–50.

[3] Feringa B. In control of motion: from molecular switches to

molecular motors. Acc Chem Res 2001;24(6):504–13.

[4] Laurent VM, Henon S, Planus E, Fodil R, Balland M, Isabey D,

Gallet F. Assessment of mechanical properties of adherent living

cells by bead micromanipulation: comparison of magnetic

twisting cytometry vs optical tweezers. J. Biomech Eng

2002;124(4):408–21.

[5] Ingber DE. Cellular basis of mechanotransduction. Biol Bull

1998;194:323–7.

[6] Freitas Jr. R. Nanomedicine. Austin: Landes Bioscience; 1999.

[7] West J, Halas NJ. Applications of nanotechnology to biotechnol-

ogy. Curr Opin Biotechnol 2000;11:215–7.

[8] Igartua M, Saulnier P, Heurtault B, Pech B, Proust JE, Pedraz JL,

Benoit JP. Development and characterization of solid lipid

nanoparticles loaded with magnetite. Int J Pharm

2002;233:149–57.

[9] Browning VM. Potential department of defense (DOD) applica-

tions of magnetic carriers. Eur Cells Mater 2002;3(Suppl. 2):5–6.

[10] Lacava ZGM, Asevedo RB, Martins LM, Freitas MLL, et al.

Biological effects of magnetic fluids: toxicity studies. J Magn

Magn Mater 1999;201(1–3):431–4.

[11] Dormer KJ. Implantable electronic otologic devices for hearing

restoration. In: Finn WE, Lopresti PG, editors. Handbook of

neuroprosthetic research methods. Boca Raton, FL: CRC Press;

2003.

[12] Austin DF. Acoustic mechanisms in middle ear sound transfer.

Otolaryngol Clin N Am 1994;27(4):641–54.

[13] Widder KJ, et al. Selective targeting of mangetic albumin

microspheres to the Yoshida sarcome: ultrstructural evaluation

of microsphere deposition. Eur J Cancer Clin Oncol

1983;19:141–7.

[14] Mornet S, Grasset F, Portier J, Dugnet E. Maghemite silica

nanoparticles for biological applications. Eur Cells Mater

2002;3(Suppl 2):110–3.

[15] Wilhelm C, Billotey C, Roger J, Pons J, Bacri J, Gazeau F.

Intracellular uptake of anionic superparamagnetic nanoparticles

as a function of their surface coating. Biomaterials 2003;24(6):

1001–11.

[16] Qhobosheane M, Santra S, Shang P, Tan W. Biochemically

functionalized silica nanoparticles. Analyst 2001;126(8):1274–8.

[17] Aliev FG, Correa-Duarte MA, Mamedov A, Ostrander JW,

Giersig M, Liz-Marzan LM, Kotov NA. Layer-by-layer assembly

of core-shell magnetite nanoparticles. Effect of silica coating on

interparticle interactions and magnetic properties. Adv Mater

1999;11(12):1006–10.

[18] Panyam J, Zhou WZ, Prabha S, Sahoo SJ, Labhasetwar V. Rapid

endo-lysosomal escape of poly(DL-lactide-co-glycolide) nanopar-

ticles: implications for drug and gene delivery. FASEB J

2002;16:1217–26.

[19] Crichton RR, Ward RJ. An overview of iron metabolism:

molecular and cellular criteria for the selection of iron chelators.

Curr Med Chem 2003;10:997–1004.

[20] Bacon BR, Stark D, Park CH, Saini S, Groman EV, Hahn PF,

Compton CC, Ferrucci Jr. JT. Ferrite particles: a new magnetic

resonance imaging contrast agent. Lack of acute or chronic

hepatotoxicity after intravenous administration. J Lab Clin Invest

1987;110(2):164–71.

[21] Massart R. Preparation of aqueous magnetic liquids in alkaline

and acidic media. Trans Magn 1981;Mag-17(2):1247–9.

[22] Akinc A, Lynn DM, Anderson DG, Langer R. Parallel synthesis

and biophysical characterization of a biodegradable polymer

library for gene delivery. J Am Chem Soc 2003;125:5316–23.

[23] Philipse AP, vanBruggen MPB, Pathmamanoharan C.

Magnetic silica dispersions: preparation and stability of surface

modified silica particles with a magnetic core. Langmuir

1994;10(1):92–9.

[24] Correa-Duarte MA, Giersig M, Kotov NA, Liz-Marzan LM.

Self-assembled monolayers of magnetite nanoparticles with and

without SiO2 coating by microwave irradiation. Langmuir

1998;14:6430–5.

[25] Wong C, Burgess JP. Modifying the surface chemistry of silica

nano-shells for immunoassays. J Young Invest 2003;6(1):1–14.

[26] Kaiser ET, Colescott RL, Bossinger CD, Cook PI. Color test for

detection of free amino groups in the solid-phase synthesis of

peptides. Anal Biochem 1970;34(2):595–8.

[27] Harris LA, Goff JD, Carmichael AY, Riffle JS, Harburn JJ,

St. Pierre TG, Saunders M. Magnetite nanoparticle

dispersions stabilized with triblock polymers. Chem Mater

2003;15:1367–77.

[28] Gan RZ, Dyer RK, Wood MW, Dormer KJ. Mass loading on

ossicles and middle ear function. Ann Otol Rhinol Laryngol

2001;110(5):478–85.

[29] Cullity BD. Elements of X-ray diffraction. 2nd ed. Reading, MA:

Addison-Wesley Publishing Inc.; 1978. p. P102.

[30] Harrison RJ, Dunin-Borkowski RE, Putnin A. Direct imaging of

nanoscale magnetic interactions in minerals. Proc Natl Acad Sci

2002;99(26):16556–61.

[31] Bergemann C, Muller-Schulte D, Oster J, Brassard LA, Lubbe

AS. Magnetite ion-enchange nano- and micro-particles for

medical, biochemical and molecular biological applications.

J Magn Magn Mater 1999;194:45–52.

[32] Sameti M, Bohr G, Ravi-Kumar MN, Kneuer C, Bakowsky M,

Nacken H, Schmidt H, Lehr CM. Stabilisation by freeze-drying of

cationically modified silica nanoparticles for gene delivery. Int J

Pharm 2003;266(1–2):51–60.

[33] Harris LA. Polymer stabilized magnetite nanoparticles. Blacks-

berg: Virginia Polytechnic Institute; 2002.

[34] Popplewell JaS L. The dependence of physical and magnetic

properties of magnetic fluids on particle size. J Magn Magn Mater

1995;149:72–8.

ARTICLE IN PRESSKenneth et al. / Biomaterials 26 (2005) 2061–20722072

[35] Panyam JaL V. Dynamics of endocytosis and exocytosis of

Poly(D,L-Lactide-co-Glycolide) nanoparticles in vascular smooth

muscle cells. Pharm Res 2003;20(2):212–20.

[36] Billotey C, Wilhelm C, Devaud M, Bacri JC, Bittoun J, Gazeau F.

Cell internalization of anionic magnemite nanoparticles: quanti-

tative effect on magnetic resonance imaging. Magn Resonance

Med 2003;49:646–54.

[37] Gan RZ, Sun Q, Dyer RK, Chang KH, Dormer KJ. Three

dimensional modeling of middle ear biomechanics and its

applications. Otol Neurotol 2002;23:271–80.

[38] Puig-De-Morales M, Grabulosa M, Alcaraz J, Mullol J, Maksym

GN, Fredberg JJ, Navajas D. Measurement of cell microrheology

by magnetic twisting cytometry with frequency domain demodu-

lation. J Appl Physiol 2001;91:1152–9.

[39] Crick FHC, Hughes AFW. The physical properties of the

cytoplasm. A study by means of the magnetic particle method.

Part 1. Exp Cell Res 1950;1:37–80.

[40] Ingber DE. Tensegrity: the architectural basis of cellular

mechanotransduction. Ann Rev Physiol 1997;59:575–99.

[41] Wang N, Naruse K, Stamenovic D, Fredberg JJ, Mijailovich SM,

Tolic-Norrelykke IM, Polte T, Mannix R, Ingber DE. Mechanical

behavior in living cells consistent with the tensegrity model. Proc

Natl Acad Sci 2001;98:7765–70.