Controlled aggregation of superparamagnetic iron oxide nanoparticles for the development of molecular magnetic resonance imaging probes

IOP PUBLISHING NANOTECHNOLOGY

Nanotechnology 19 (2008) 265102 (7pp) doi:10.1088/0957-4484/19/26/265102

Controlled aggregation ofsuperparamagnetic iron oxidenanoparticles for the development ofmolecular magnetic resonance imagingprobesB A Larsen1, M A Haag1, N J Serkova2, K R Shroyer3 andC R Stoldt1

1 Department of Mechanical Engineering, University of Colorado, Boulder,CO 80309-0427, USA2 Department of Anesthesiology, Biomedical MRI/MRS Cancer Center Core,University of Colorado at Denver and Health Sciences Center, Aurora, CO 80045, USA3 Department of Pathology, University of Colorado at Denver and Health Sciences Center,Aurora, CO 80045, USA

E-mail: [email protected]

Received 19 December 2007, in final form 20 March 2008Published 19 May 2008Online at stacks.iop.org/Nano/19/265102

AbstractA method for synthesizing superparamagnetic iron oxide (SPIO) multi-nanoparticle aggregatesas molecular magnetic resonance imaging (MRI) contrast agents is described. The approachutilizes organic acid/base interactions in the colloid to induce highly controllable nanoparticleaggregation. Monodisperse aggregates with diameters as large as 100 nm are synthesized bymanipulating the interfacial surface chemistry of the SPIO nanoparticles in tetrahydrofuransolvent. Subsequent phospholipid micelle encapsulation yields micellar multi-SPIO (mmSPIO)aggregates with enhanced T2 relaxivity (368.0 s−1 mmol−1 Fe) as compared to micellar singleparticle SPIO (302.0 s−1 mmol−1 Fe). mmSPIO conjugated to anti-CA125 monoclonalantibodies were incubated with ovarian carcinoma cell lines to demonstrate targeted in vitromolecular MRI, resulting in a 66% shortening in T2 time for CA125 positive NIH:OVCAR-3cells and a less than 3% change in T2 time for CA125 negative SK-OV-3 cells. The controllableaggregation of mmSPIO shows potential for the development of molecular MRI contrast agentswith optimal sizes for specific diagnostic imaging applications.

1. Introduction

Superparamagnetic iron oxide (SPIO) nanoparticles areprevalent in molecular imaging as magnetic resonance imaging(MRI) contrast agents due to their localized shortening ofspin–spin (T2) proton relaxation times [1–5]. Currently,there is wide interest in using targeted SPIO nanoparticlesin diagnostic imaging applications, with a number ofrecent studies demonstrating the clinical potential for cancerdiagnosis and tumor detection. By functionalizing SPIOnanoparticles with targeting vectors (i.e. antibodies, small

molecular ligands, aptamers), known biomarkers for cancer(e.g. cell surface onco-proteins) can be specifically bound,yielding local T2 contrast of target-positive tumors [6–10]. Inaddition to biofunctionalization, SPIO nanoparticle materialproperties can be manipulated by changes in size and surfacechemistry. These approaches enable ‘tuning’ of the colloidalmagnetic properties [11]. Specifically, the local T2 protonrelaxivity of SPIO nanoparticles can be modified via interfacialsurface chemistry to induce aggregation or dispersion in thecolloid, thus providing control of the ultimate nanoparticleaggregate size and magnetic properties, with the end goal

0957-4484/08/265102+07$30.00 © 2008 IOP Publishing Ltd Printed in the UK1

Nanotechnology 19 (2008) 265102 B A Larsen et al

Figure 1. Schematic showing the SPIO aggregation and encapsulation procedure. (A) SPIO nanoparticles dispersed in THF. (B) Acetic acid isintroduced, partially displacing heptanoic acid from the SPIO surface. (C) Aggregation of SPIO occurs due to decreased steric hindrance ofthe SPIO surface. (D) The resulting aggregates are dried on a film of phospholipids. (E) The aggregates and phospholipids are resuspended inaqueous solution, encapsulating the aggregates to yield mmSPIO.

being the optimization of the colloid for a specific imagingapplication.

T2 contrast enhancement from SPIO nanoparticle aggre-gation is a well-established phenomenon. Weissleder andcoworkers have exploited the measurable shortening of T2 re-laxation time to develop ‘magnetic relaxation switches’ forsensing various molecular interactions [12–14]. In our pre-vious work, T2 relaxation times were increasingly shortenedwith greater levels of aggregation induced by DNA crosslinkedSPIO nanoparticles [15]. In addition, SPIO nanoparticle ag-gregates have been recently studied as MRI contrast agents.One such example involves a polymer encapsulation of SPIOnanoparticles, utilizing a polyacrylamide matrix to maintainand functionalize the aggregates [16]. Another example uti-lizes amphiphilic block copolymers to control SPIO nanoparti-cle aggregation, yielding aggregates with enhanced propertiesfor T2-weighted MRI [17]. Both examples involved significantpolymer chemistry and neither offered fine control of aggregatesize.

This study reports a new synthetic approach to thepreparation of encapsulated SPIO nanoparticle aggregates forT2 MRI using a simple and highly controllable procedure.Specifically, a SPIO nanoparticle aggregation method isdescribed here that utilizes acid/base interactions involvingSPIO nanoparticles in organic solvent. By increasing theacidity of the organic solution, SPIO nanoparticles can beincrementally aggregated to a desired size and encapsulatedin a phospholipid monolayer to yield stable micelles of SPIOnanoparticle aggregates. Consistent with the aforementionedstudies, these micellar multi-SPIO (mmSPIO) aggregatesdemonstrate significantly shorter T2 relaxation times thansingle particle contrast agents. The precise control affordedby this organic acid mediated method enables synthesisof nanoparticle aggregates optimized for size and magneticproperties. The efficacy of mmSPIO as a molecular MRIcontrast agent is demonstrated by an in vitro molecular MRIstudy targeting cancer antigen 125 (CA125), a glycoproteincommonly expressed in ovarian surface epithelial carcinomacells [18].

2. Experimental details

2.1. SPIO synthesis and characterization

The SPIO nanoparticles used in this study were synthesized bythe solvothermal processing route, and further details on thematerial synthesis, characterization, and magnetic propertiesof the Fe3O4 product are described elsewhere [19]. Theas-synthesized SPIO nanoparticles are nearly monodisperse(diameters ranging from 10–14 nm) with a hydrophobicheptanoic acid surface termination.

2.2. SPIO aggregation and encapsulation

SPIO nanoparticles in toluene were resuspended in tetrahydro-furan (THF) and titrated with a 1% solution of acetic acid untilthe desired level of aggregation was reached. SPIO aggregatesize and size distribution was measured by dynamic light scat-tering (DLS) using a Brookhaven BI-200SM Goniometer. Toeliminate dust from the sample, SPIO nanoparticles in THFwere filtered through 200 nm syringe filters into a glass cu-vette for DLS. Measurements were made in triplicate at roomtemperature for 3 min with the detector at 90◦ and size distribu-tions were obtained using a CONTIN algorithm. SPIO singleparticle and aggregate samples were drop cast from THF sol-vent onto a TEM grid prior to imaging on a Philips CM10 TEMoperating at 80 kV.

An overview of the mmSPIO synthesis is illus-trated in figure 1. This method utilizes hydropho-bic/hydrophilic interactions between the amphiphilic phos-pholipids and the hydrophobic nanoparticle surface in aque-ous solution for micellar encapsulation. Forty mg ofSPIO nanoparticle aggregates were added to a film ofdried PEGylated phospholipids from Avanti Polar lipids(20 mg 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide(polyethyleneglycol)2000] and 2 mg 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N -[carboxyfluorescein])and dried with inert gas in a round bottom flask in a processsimilar to previous studies [20, 21]. Vacuum was applied to thedried SPIO nanoparticle aggregate and phospholipid film for

2

Nanotechnology 19 (2008) 265102 B A Larsen et al

1 h. The mixture was then resuspended with 1 ml deionizedwater by bath sonication for 5 min at 60 ◦C to yield mmSPIO.The phospholipids and resulting mmSPIO solution were pro-tected from light during and after the encapsulation process toavoid photo-bleaching the fluorescein. After one month, mmS-PIO in deionized water remained stable and exhibited no SPIOdegradation as determined by DLS, indicating effective stabi-lization via phospholipid micelle encapsulation.

Single particle micellar SPIO and mmSPIO aggregatesamples were divided in three dilutions and 10 µl sampleswere prepared in sealed microcapillary tubes. A Carr–Purcell–Meiboom–Gill pulse sequence was used to determine the T2

relaxation time. Molar relaxation rates were calculated fromthe linear slope of the inverse T2(1/T2) relaxation times andFe3O4 molarity, as determined by inductively coupled plasmaoptical emission spectrometry (ICP-OES) on an ARL 3410+.The linear fits of the T2 data demonstrate near ideal correlation(R2 > 0.99).

2.3. mmSPIO crosslinking to monoclonal antibodies

mmSPIO were crosslinked to monoclonal antibodies by form-ing a thioether bond between the maleimide functionalizedphospholipids on the mmSPIO and sulfhydryl groups on themonoclonal antibodies. 500 µg of IgG1 anti-CA125 mon-oclonal antibodies (Mab) from Biodesign (clone X306) and500 µg of rat serum IgG from Sigma Aldrich were separatelyreconcentrated in reduction buffer (PBS—0.1 M sodium phos-phate, 0.15 M NaCl, 10 mM EDTA, pH 7.2.) using Milli-pore Amicon Ultracel 100k centrifugal filters to a concentra-tion of 1 mg ml−1. The antibody solutions were separatelyadded to 6 mg of 2-mercaptoethylamine· HCl (2-MEA, Pierce)and allowed to react for 90 min at 37 ◦C, reducing the Mabheavy chain disulfide bonds to yield free sulfhydryl functionalgroups. The reduced antibody solutions were desalted sep-arately with a Zeba spin desalt column using a conjugationbuffer (PBS—0.05 M sodium phosphate, 0.15 M NaCl, 10 mMEDTA, pH 7.2) to prevent reformation of disulfide bonds dur-ing the crosslinking to the mmSPIO.

40 mg of mmSPIO in 1 ml of deionized water waswashed by centrifugation with 1 ml of the previouslymentioned conjugation buffer to remove unfilled phospholipidmicelles resulting from excess phospholipids during theencapsulation process. The mmSPIO in conjugation bufferwas divided into two 10 mg quantities and added to separateround bottom flask with the reduced antibody solutionssolution. The mmSPIO/Mab solution was reacted overnightat room temperature with gentle agitation on an orbitalmixer. The resulting CA125 specific mmSPIO (anti-CA125mAb conjugate) and non-specific mmSPIO (Rat serum IgGconjugate) were separately washed by centrifugation with 1 mlstorage buffer (PBS—0.05 M sodium phosphate, 0.15 M NaCl,pH 7.2) for in vitro use.

2.4. In vitro molecular MRI

An in vitro molecular MRI study was performed usingtwo human ovarian carcinoma cell lines, NIH:OVCAR-3 (CA125 positive) and SK-OV-3 (CA125 negative) from

American Type Culture Collection (ATCC). The cell lines werecultured in RPMI 1640 (OVCAR-3) and McCoy’s 5A medium(supplemented with 10–20% fetal bovine serum) until theyreached 70% confluency in 75 cm2 cell culture flasks. Totalnumber of cells was between 8 and 11 million cells.

1 ml of CA125 specific mmSPIO and 1 ml non-specificmmSPIO in storage buffer were diluted with cell culturemedium to yield two 60 ml volumes of mmSPIO/mediasolutions. Ten ml per flask were then added to 6 cultureflasks of OVCAR-3 cells and 6 culture flasks of SK-OV-3cells, providing a positive and negative control of each cellline for CA125 specific mmSPIO and non-specific mmSPIO.Two culture flasks of each cell line were incubated with bufferonly (without mmSPIO) to provide baseline samples for flowcytometry and MRI. After incubation for 30 min at 4 ◦C, eachflask was washed with 10 ml of PBS to remove unboundmmSPIO and cells were detached with 5 ml of 0.25% trypsinsolution. Upon cell detachment, the trypsin solution wasquenched with 10 ml of medium and the cells were countedwith a hemocytometer. Media and trypsin were removed fromthe cells by washing with PBS.

In vitro T2-weighted MRI of cell pellets (after centrifu-gation) was performed using a 4.7 T Bruker PharmaScan tocalculate T2 values and verify CA125 binding of the CA125specific mmSPIO. Briefly, the microcentrifuge tubes, each con-taining cell pellets and PBS from one 75 cm2 flask, were placedonto the custom designed holder and inserted into a Brukervolume coil (62 mm diameter). The coil, tuned to the 1H fre-quency of 200 MHz, was used for radiofrequency (RF) trans-mission and receiving. A sequence of multi-slice multi-echo(MSME) T2-weighted scans were performed with various echotimes (TE). The following parameters were used: field of view(FOV) 6.0 cm, slice thickness 2 mm, with 2 axial or sagittalslices, matrix size 128 × 256, relaxation time TR = 2000 ms,TE = 12/24/36/48/60 etc ms, total number of echoes 12,number of averages‘1, total scan time 10 min 8 s. The cal-culations of T2 values were performed automatically for se-lected region of interest (ROI) using Bruker ParaVision soft-ware (Version 3.0, Bruker Medical, Billerica, MA) accordingto equation (1), where S is a MRI signal intensity and C2 = M0

(1−e−R/T 1) is a constant. In addition, flow cytometry was per-formed on a BD Biosciences FACScalibur to detect fluoresceinstaining by the fluorescent phospholipids used in mmSPIO en-capsulation.

S = C2e−TE/T 2. (1)

3. Results

3.1. Characterization of single particle micellar SPIO andmmSPIO

SPIO nanoparticle aggregation is achieved by titratingthe as-synthesized colloid (heptanoic acid terminated SPIOnanoparticles in THF) with acetic acid. For example, additionof a 1% (vol/vol) solution of glacial acetic acid to the colloidinduced incremental aggregation of the SPIO nanoparticles perthe tabulated hydrodynamic diameters determined by DLS intable 1. The aggregation is precisely controlled with a 1%

3

Nanotechnology 19 (2008) 265102 B A Larsen et al

Table 1. Tabulated hydrodynamic diameters and polydispersityindices obtained by DLS of SPIO nanoparticle aggregation in THFduring titration with acetic acid.

% of acetic acid inTHF (vol/vol)

Aggregatehydrodynamicdiameter (nm)

Polydispersityindex

0.00% 15.7 0.0030.23% ± 0.04% 31.4 0.0120.33% ± 0.03% 52.2 0.0010.41% ± 0.03% 91.1 0.6990.47% ± 0.03% 100.5 0.372

solution of acetic acid and finer control is possible by furtherdilution of the titrant. The low polydispersity of hydrodynamicdiameters in table 1 for aggregates as large as 50 nm is asignificant result, indicating homogeneous properties of theSPIO aggregates due to high colloid monodispersity. Infigure 2(A), a TEM image representative of dispersed SPIOnanoparticles is shown. In figures 2(B) and (C), TEM imagesshow SPIO nanoparticles after increasing amounts of aceticacid are added, demonstrating a range of achievable aggregatediameters prior to phospholipid encapsulation. Note that theSPIO nanoparticle aggregates shown in figure 2 are drop caston a TEM grid prior to imaging, resulting in some distortionof the observed shape and causing the observed diameter ofthe aggregates in (B) and (C) to appear slightly larger than thehydrodynamic diameter measured in suspension by DLS.

T2 relaxometry results comparing the relaxation ratessingle particle micellar SPIO and mmSPIO aggregates areshown in figure 3. The single particle sample yields aT2 relaxation rate of 302.7 s−1 mmol−1 Fe (figure 3(a)),while mmSPIO samples having 24 and 40 particles peraggregate on average give T2 relaxation rates of 325.8 and368.0 s−1 mmol−1 Fe (figures 3(b) and (c)), respectively.As demonstrated by the results in figure 3, the 40 particlemmSPIO aggregate yields an approximately 20% higher molarT2 relaxation rate than single nanoparticle micellar SPIO,consistent with the conclusions of previous multi-particlestudies [16, 17, 22].

3.2. MRI of ovarian carcinoma cells

In figure 4(A), MSME T2 weighted MRI of NIH:OVCAR-3 cell pellets indicate CA125 binding of the CA125 specificmmSPIO. The cell pellet of OVCAR-3 cells incubatedwith CA125 specific mmSPIO shows enhanced negativeT2 contrast compared to the cell pellet of OVCAR-3cells incubated with non-specific mmSPIO and, moreover,without mmSPIO. Figure 4(B) shows the significant contrastdifference between OVCAR-3 cell pellets incubated withCA125 specific mmSPIO and non-specific mmSPIO comparedto the negligible contrast difference between the SK-OV-3 cellpellets under the same incubation conditions. The calculatedT2 times are shown in table 2 for OVCAR-3 and SK-OV-3 cells incubated without mmSPIO, with CA125 specificmmSPIO, and with non-specific mmSPIO. From table 2, theT2 percentage change between the CA125 specific and non-specific mmSPIO aggregates of the OVCAR-3 cell line is

Figure 2. TEM images of (A) 12 nm SPIO nanoparticles prior toaggregation, (B) ∼60 nm and (C) ∼100 nm SPIO nanoparticleaggregates following the controlled addition of acetic acid.Images (A)–(C) are shown prior to phospholipid encapsulationand (B)–(C) represent a range of aggregate sizes achievable with acidtitration in THF.

∼66%, while the T2 percentage change between the CA125specific and non-specific mmSPIO aggregates of the SK-OV-3 cell line is less than 3%. Due to the slight differencesin cell volumes, the OVCAR-3 cell pellets contained anaverage of 10.5 million cells and the SK-OV-3 cell pellets

4

Nanotechnology 19 (2008) 265102 B A Larsen et al

(B)

(A)

(C)

Figure 3. T2 relaxometry results plotted as mmol of Fe versus 1/T2

(s−1) for aqueous samples of (A) single particle micellar SPIO,(B) 24 particle mmSPIO aggregates, and (C) 40 particle mmSPIOaggregates. Molar relaxation rates for each sample are determinedfrom the linear fit of the relaxation data.

8.0 million cells. CA125 binding is further supported bythe flow cytometry histograms shown in figure 5. Strongfluorescein staining of OVCAR-3 cells (CA125 positive)and weak staining of SK-OV-3 cells (CA125 negative) isdemonstrated. The strong increase in fluorescein intensity inthe NIH:OVCAR-3 cells is due to the fluorescein-phospholipidencapsulation of the CA125 specific mmSPIO.

4. Discussion

By titrating an SPIO/THF colloid with acetic acid, wedemonstrate a facile method for the precise control ofthe nanoparticle aggregate size via Lewis acid/Lewis basechemistry prior to encapsulation in the phospholipid micellefor molecular MRI. The observed aggregation of SPIOnanoparticles in THF upon addition of acetic acid is

Figure 4. In vitro molecular T2-weighted MRI of (A) the CA125positive ovarian carcinoma cell line OVCAR-3 incubated withCA125 specific mmSPIO, with non-specific mmSPIO, and withbuffer without mmSPIO; and (B) the OVCAR-3 and the CA125negative ovarian carcinoma cell line SK-OV-3 incubated with CA125specific mmSPIO and with non-specific mmSPIO.

Figure 5. Flow cytometry data for SK-OV-3 and OVCAR-3 cellsincubated without mmSPIO ((A) and (C)) and with CA125 specificmmSPIO ((B) and (D)).

(This figure is in colour only in the electronic version)

consistent with published colloid and surface chemistryresearch. Aggregation and dispersion in the colloid ismediated by the adsorption of the acidic surfactant at the SPIOnanoparticle surface. An early study by Chen et al investigatedsolvent effects on iron nanoparticle dispersions [23], whereincreased dispersion of iron nanoparticles was reported upondeprotonation of the acidic surfactant in THF. Deprotonationfacilitates adsorption of the surfactant’s carboxylate grouponto the nanoparticle surface, thus stabilizing the colloid anddispersing the nanoparticle inclusions.

5

Nanotechnology 19 (2008) 265102 B A Larsen et al

Table 2. Tabulated T2 times determined by MSME T2-weightedMRI for OVCAR-3 and SK-OV-3 cells incubated without SPIO, withnon-specific SPIO, and with CA125 specific SPIO.

OVCAR-3 SK-OV-3Incubation Cell pellet (ms) Cell pellet (ms)

Without SPIO 158.2 ± 48.3 242.5 ± 25.7With non-specific SPIO 129.0 ± 56.3 116.7 ± 11.8a

With CA125 specific SPIO 47.7 ± 11.3a,b 120.0 ± 21.2a

a p < 0.01 compared to control.b p < 0.05 compared to mmSPIO.

In another study, Krishnakumar and Somasundaraninvestigated surfactant adsorption on metal oxide surfaces inorganic solvents and reported decreased surfactant adsorptionwith increased solvent polarity, resulting from the competitiveadsorption of solvent and surfactant at the oxide surface [24].In our case, with the addition of the more polar acetic acid toTHF, the overall colloid solvent polarity is increased resultingin exchange of heptanoic acid for acetic acid at the SPIOnanoparticle surface. With removal of the significantly largersurfactant species, the colloid is destabilized and nanoparticleaggregation results.

By displacing the heptanoic acid during titration, particlecollisions are more likely to result in aggregates due tothe decreased steric hindrance between nanoparticles in thecolloid. Based on the short timescale of the observedaggregation, the aggregation mechanism is diffusion limited.The diffusion limited aggregation regime is characterized bythe formation of fractal aggregates having decreased fractaldimension with increasing aggregate size, which explains theincreasing polydispersity for the larger nanoparticle aggregates(see table 1). A similar broadening of size distribution has beenreported for aggregation of ∼20 nm silica nanoparticles [25].The dense aggregate structures in figures 2(b) and (c) isthe result of the fractal aggregate restructuring to a higherfractal dimension, consistent with observations of similaraggregates [26].

The enhanced T2 relaxation of mmSPIO measured hereis significant when combined with the highly controllableaggregation process also demonstrated. The enhanced T2

relaxation of mmSPIO is in agreement with previous studies ofaggregated SPIO [12–14, 16, 17, 22]. Specifically, mmSPIOsamples possess higher relaxivity than SPIO due to theincreased proton spin dephasing from the cooperating SPIOcores in the aggregates [13]. This cooperative T2 relaxationrate enhancement of aggregated SPIO cores is explained by atheoretical model by Roch et al which adapts the conventionalSPIO model by accounting for the greater magnetic momentand longer correlation time caused by aggregation [22]. Theincreased magnetic moment and radius of mmSPIO enhancethe T2 relaxation rate (R2) without affecting the T1 relaxationrate (R1), thus increasing R2/R1 ratio and improving thesignal-to-noise ratio of mmSPIO as a T2 contrast agent [22].

The facile size control of mmSPIO is significant to in vivomolecular MRI applications. This ‘post-processing’ approachenables a range of SPIO nanoparticle aggregates to be preparedfrom a single SPIO nanoparticle synthesis reaction and offers

the advantage of controlling the aggregate size in a separatestep before encapsulation. This ‘bottom-up’ approach enablescontrolled aggregation until a desired size is reached, ratherthan fixed aggregate sizes synthesized by the simultaneousaggregation/encapsulation of other methods [16, 17, 27], Asexpected, pharmacokinetic/pharmacodynamic studies of SPIOand other sub-micron colloids report a strong relationshipbetween particle size and biodistribution [5, 28–32]. mmSPIOsizing via controlled aggregation provides a route to sizeoptimization of the T2 contrast agent for an intended in vivotarget without altering the original SPIO synthesis procedure.The as-synthesized single particle micellar SPIO is wellsuited for blood-pool imaging, while the 55 nm mmSPIOaggregates prepared by the synthetic route described hereare, for example, well suited for liver reticuloendothelialsystem imaging [30]. New molecular MRI applications mayinclude imaging of micrometastatic ovarian cancer, exploitingcontrolled mmSPIO aggregation to optimally size the contrastagent for intraperitoneal delivery and peritoneal implantbinding.

5. Conclusions

In conclusion, colloid interfacial surface chemistry ismanipulated to yield monodisperse mmSPIO with enhancedT2 relaxivity compared to as-synthesized single particle SPIO.Precise chemical control of nanoparticle aggregation is used tosynthesize mmSPIO as large as 100 nm, enabling optimizationof mmSPIO for specific in vivo applications based on sizerelated biodistribution. Finally, binding of CA125 specificmmSPIO toward CA125 on ovarian carcinoma cells is shown,thus demonstrating viability of mmSPIO as an optimal T2-weighted molecular MRI contrast agent. The syntheticapproach described here benefits molecular imaging researchby enabling post-processing of SPIO to achieve T2 relaxationenhancement and size control.

Acknowledgments

The Cancer League of Colorado and the University ofColorado Cancer Center (P30 CA046934 Core grant) areacknowledged for their support of this study. The authors aregrateful to Jaimi Brown and Kristen Scaff from the Universityof Colorado at Denver and Health Sciences Center for cellculture and in vitro preparation, and Dr Richard Shoemakerfrom the Department of Chemistry and Biochemistry at theUniversity of Colorado, Boulder for NMR assistance.

References

[1] Jaffer F A and Weissleder R 2005 Molecular imaging in theclinical arena Jama-J. Am. Med. Assoc. 293 855–62

[2] Sosnovik D E, Nahrendorf M and Weissleder R 2007Molecular magnetic resonance imaging in cardiovascularmedicine Circulation 115 2076–86

[3] Weissleder R and Mahmood U 2001 Molecular imagingRadiology 219 316–33

6

Nanotechnology 19 (2008) 265102 B A Larsen et al

[4] Wu E X, Tang H Y and Jensen J H 2004 Applications ofultrasmall superparamagnetic iron oxide contrast agents inthe MR study of animal models NMR Biomed. 17 478–83

[5] Duguet E, Vasseur S, Mornet S and Devoisselle J M 2006Magnetic nanoparticles and their applications in medicineNanomedicine 1 157–68

[6] Katz E and Willner I 2004 Integrated nanoparticle–biomoleculehybrid systems: Synthesis, properties, and applicationsAngew. Chem. Int. Edn 43 6042–108

[7] Sun E Y, Josephson L, Kelly K A and Weissleder R 2006Development of nanoparticle libraries for biosensingBioconjug. Chem. 17 109–13

[8] Sunderland C J, Steiert M, Talmadge J E, Derfus A M andBarry S E 2006 Targeted nanoparticles for detecting andtreating cancer Drug Dev. Res. 67 70–93

[9] Thorek D L J, Chen A, Czupryna J and Tsourkas A 2006Superparamagnetic iron oxide nanoparticle probes formolecular imaging Ann. Biomed. Eng. 34 23–38

[10] Weissleder R 2006 Molecular imaging in cancerScience 312 1168–71

[11] Jun Y W et al 2005 Nanoscale size effect of magneticnanocrystals and their utilization for cancer diagnosis viamagnetic resonance imaging J. Am. Chem. Soc. 127 5732–3

[12] Tsourkas A, Hofstetter O, Hofstetter H, Weissleder R andJosephson L 2004 Magnetic relaxation switchimmunosensors detect enantiomeric impuritiesAngew. Chem. Int. Edn 43 2395–9

[13] Perez J M, Josephson L, O’Loughlin T, Hogemann D andWeissleder R 2002 Magnetic relaxation switches capable ofsensing molecular interactions Nat. Biotechnol.20 816–20

[14] Josephson L, Perez J M and Weissleder R 2001 Magneticnanosensors for the detection of oligonucleotide sequencesAngew. Chem. Int. Edn 40 3204

[15] Larsen B A, Haag M A, B Stowell M H, Walther D C,Pisano A P and Stoldt C R 2007 Controlling nanoparticleaggregation in colloidal microwave absorbers via interfacechemistry Proc. SPIE 652519

[16] Moffat B A, Reddy G R, McConville P, Hall D E,Chenevert T L, Kopelman R R, Philbert M, Weissleder R,Rehemtulla A and Ross B D 2003 A novel polyacrylamidemagnetic nanoparticle contrast agent for molecular imagingusing MRI Mol. Imag. 2 324–32

[17] Berret J F, Schonbeck N, Gazeau F, El Kharrat D, Sandre O,Vacher A and Airiau M 2006 Controlled clustering ofsuperparamagnetic nanoparticles using block copolymers:design of new contrast agents for magnetic resonanceimaging J. Am. Chem. Soc. 128 1755–61

[18] Leake J, Woolas R P, Daniel J, Oram D H and Brown C L 1994Immunocytochemical and serological expression of Ca125-a clinicopathological study of 40 malignant ovarianepithelial tumors Histopathology 24 57–64

[19] Barker A J, Cage B, Russek S and Stoldt C R 2005 Ripeningduring magnetite nanoparticle synthesis: resulting interfacialdefects and magnetic properties J. Appl. Phys. 98 063528

[20] Nitin N, LaConte L E W, Zurkiya O, Hu X and Bao G 2004Functionalization and peptide-based delivery of magneticnanoparticles as an intracellular MRI contrast agent J. Biol.Inorg. Chem. 9 706–12

[21] Dubertret B, Skourides P, Norris D J, Noireaux V,Brivanlou A H and Libchaber A 2002 In vivo imaging ofquantum dots encapsulated in phospholipid micellesScience 298 1759–62

[22] Roch A, Gossuin Y, Muller R N and Gillis P 2005Superparamagnetic colloid suspensions: water magneticrelaxation and clustering J. Magn. Magn. Mater. 293 532–9

[23] Chen W J, Wong S S, Peng W G and Wu C D 1991 Effect ofacid–base interaction on magnetic dispersion containingalpha-fe metal particles IEEE Trans. Magn. 27 4648–50

[24] Krishnakumar S and Somasundaran P 1994 Role of surfactantadsorbent acidity and solvent polarity inadsorption–desorption and surfactants from nonaqueousmedia Langmuir 10 2786–9

[25] Schlomach J and Kind M 2004 Investigations on the semi-batchprecipitation of silica J. Colloid Interface Sci. 277 316–26

[26] Aubert C and Cannell D S 1986 Restructuring of colloidalsilica aggregates Phys. Rev. Lett. 56 738

[27] Ai H, Flask C, Weinberg B, Shuai X, Pagel M D, Farrell D,Duerk J and Gao J M 2005 Magnetite-loaded polymericmicelles as ultrasensitive magnetic-resonance probesAdv. Mater. 17 1949

[28] Corot C, Robert P, Idee J-M and Port M 2006 Recent advancesin iron oxide nanocrystal technology for medical imagingAdv. Drug Deliv. Rev. 58 1471–504

[29] Mornet S, Vasseur S, Grasset F and Duguet E 2004 Magneticnanoparticle design for medical diagnosis and therapyJ. Mater. Chem. 14 2161–75

[30] Okuhata Y 1999 Delivery of diagnostic agents for magneticresonance imaging Adv. Drug Deliv. Rev. 37 121–37

[31] Chouly C, Pouliquen D, Lucet I, Jeune J J and Jallet P 1996Development of superparamagnetic nanoparticles for MRI:effect of particle size, charge and surface nature onbiodistribution J. Microencapsul. 13 245–55

[32] Stolnik S, Illum L and Davis S S 1995 Long circulatingmicroparticulate drug carriers Adv. Drug Deliv. Rev.16 195–214

7