Protease-Specific Nanosensors for Magnetic Resonance Imaging

Eyk Schellenberger,* Franziska Rudloff, Carsten Warmuth, Matthias Taupitz, Bernd Hamm, and Jorg Schnorr

Department of Radiology, Charite - Universitatsmedizin Berlin; Chariteplatz 1; 10117 Berlin, Germany. Received August 1, 2008;Revised Manuscript Received October 27, 2008

Imaging of enzyme activity is a central goal of molecular imaging. With the introduction of fluorescent smartprobes, optical imaging has become the modality of choice for experimental in ViVo detection of enzyme activity.Here, we present a novel high-relaxivity nanosensor that is suitable for in ViVo imaging of protease activity bymagnetic resonance imaging. Upon specific protease cleavage, the nanoparticles rapidly switch from a stablelow-relaxivity stealth state to become adhesive, aggregating high-relaxivity particles. To demonstrate the principle,we chose a cleavage motif of matrix metalloproteinase 9 (MMP-9), an enzyme important in inflammation,atherosclerosis, tumor progression, and many other diseases with alterations of the extracellular matrix. On thebasis of clinically tested very small iron oxide particles (VSOP), the MMP-9-activatable protease-specific ironoxide particles (PSOP) have a hydrodynamic diameter of only 25 nm. PSOP are rapidly activated, resulting inaggregation and increased T2*-relaxivity.

INTRODUCTION

The advent of fluorescent smart sensor probes activatable byproteases ignited the development of reporter probes (1) andinstrumentation (2) establishing optical imaging as the modalityof choice for in ViVo detection of enzyme activity (3).

To take advantage of the high resolution and deep-tissueimaging capabilities of magnetic resonance imaging (MRI),suitable for whole-body imaging of humans, as well as theabsence of radiation exposure, several new probe designs wereintroduced. Meade and co-workers developed a reporter probeactivatable by �-galactosidase on the basis of a paramagneticcontrast agent and demonstrated its successful use for thevisualization of gene transfer by MRI (4).

A paramagnetic polymerization probe was presented byBogdanov et al., which allowed imaging of tissue peroxidaseactivity by MRI (5). Josephson and co-workers introducedmagnetic relaxation switches sensing enzymatic activity basedon superparamagnetic high-relaxivity probes for in Vitro MRI(6). Upon protease activation, these sterically stabilized cross-linked iron oxide particles (CLIO) switch from large clusteredhigh-relaxivity aggregates to smaller low-relaxivity singleparticles. This technique is not suitable for in ViVo application,since the nonactivated clustered particles are too large to havea favorable bioavailability and, more importantly, activation byelevated enzyme activity in the target tissue reduces therelaxivity and consequently the contrast effect.

Here, we describe a new design of protease-sensitive nanosen-sors, which are based on electrostatically stabilized, citrate-coated very small iron oxide particles (VSOP) with a hydro-dynamic diameter of 7.7 ( 2.1 nm (Figure 2a) instead of largersterically stabilized nanoparticles (over 20 nm). We havepreviously developed a pharmacologically formulated variantof these superparamagnetic particles (VSOP-C184) for magneticresonance angiography, which was successfully tested in aclinical phase 1 trial and currently undergoing clinical phase 2testing (7).

As cleavage motif, we chose a recognition site for MMP-9and MMP-2, which was previously used to prepare a MMP-9-activatable fluorescent probe (8, 9). MMPs have been identified

as important regulators of pathological remodeling of extracel-lular matrix in cancer and inflammation (10). Using fluorescentsmart probes, Bremer et al. developed a new concept to monitorthe success of MMP inhibition in cancer by in ViVo imaging,which may ultimately enable monitoring of therapeutic responsesin the large spectrum of diseases with alterations of theextracellular matrix (11).

EXPERIMENTAL PROCEDURES

Synthesis. (1) The synthesized fluorescein-labeled MMP-9peptide NH2-GGPRQITAG-K(FITC)-GGGG-RRRRR-G-RRRRR-amide (the italicized amino acids correspond to the MMP-9substrate (9)) was confirmed by MALDI-TOF (MALDI 2,Shimadzu). The molecular weight was within 1 Da of theexpected value. 1 mg of the MMP-9 peptide (in 100 µL DMSO)was reacted with 31 mg NHS-mPEG (O-[(N-succinimidyl)suc-cinyl-aminoethyl]-O′-methyl-poly(ethylene glycol) 5000, Fluka)in 300 µL 0.1 M HEPES, pH 7.5, overnight at 4 °C. Theresulting MMP-9-peptide-mPEG copolymer was purified by gelfiltration with BioGel P6 (BioRad, in 10 mM HEPES, 140 mMNaCl, pH 7.5). The absence of unreacted MMP-9 peptide wasconfirmed by HPLC using a reverse-phase C18 column withacetonitrile/water gradient. The concentration of the peptide-mPEG copolymers was determined spectrophotometrically bymeasuring peptide-bound fluorescein absorption (absorption at494 nm, extinction coefficient of 72 000 M-1 cm-1).

(2) VSOP (VSOP-C200) was provided by Ferropharm GmbH.To prepare 6×-MMP-9-PSOP (with six peptide-mPEG copoly-mers per VSOP), 1.25 nmol VSOP (3.25 µmol Fe assuming2600 iron atoms per particle, data from Ferropharm GmbH) in250 µL HEPES buffer (10 mM HEPES, 140 mM NaCl, pH7.5) were mixed with 7.5 nmol peptide-mPEG in 250 µL HEPESbuffer and stirred immediately. Other ratios (3 to 16) wereprepared accordingly. Hydrodynamic diameters of MMP-9-PSOPs were determined by dynamic light scattering (ZetasizerNano ZS, Malvern Instruments) in the same buffer. Relaxivitieswere measured with an MR spectrometer at 0.94 T (Bruker,Minispec MQ 40) according to the manufacturer’s instructions.

For quenching experiments (Figure 3), 50 pmol VSOP wasmixed with different amounts of the peptide-mPEG copolymers(150 pmol to 800 pmol) in 200 µL and subjected to fluorescence* E-mail: eyk.schellenberger@charite.de.

Bioconjugate Chem. 2008, 19, 2440–24452440

10.1021/bc800330k CCC: $40.75 2008 American Chemical SocietyPublished on Web 11/14/2008

measurements (excitation/emission 494/521 nm, Hitachi fluo-rescence spectrophotometer F-7000). Ratios of these quenchedfluorescences and corresponding unquenched fluorescences ofequal copolymer concentrations without VSOP addition werecalculated and compared with the fluorescence of free fluores-cein dye mixed with VSOP accordingly.

Enzyme Activation. To demonstrate the ability of MMP-9to activate MMP-9-PSOP (0.125 nmol particles in 50 µL),hydrodynamic size measurements were carried out before andafter adding MMP-9 (7.4 U MMP-9, human, Calbiochem) at37 °C. For experiments with enzyme inhibition, MMP-2/MMP-9Inhibitor II (30 µM final concentration, Calbiochem) was addedbefore the addition of MMP-9 enzyme. The hydrodynamicdiameters (Zetasizer) were monitored for 2 h. The incubationbuffer for enzyme experiments was 10 mM HEPES, 140 mMNaCl, 1.3 mM CaCl2, and 50 µM ZnCl2, pH 7.5.

MR Imaging. For the MR experiments (Figure 4b), fourdifferent particle concentrations of 6×-MMP-9-PSOP (300 nM,150 nM, 75 nM, and 38 nM particle concentration) wereincubated with 1.3 U MMP-9 in 200 µL incubation buffer (seeabove) at 37 °C. All inhibitor experiments (Figure 4c,d) weredone under identical conditions using a particle concentrationof 300 nM. All enzyme reactions were imaged with a clinical1.5-T MR scanner (Siemens Sonata). The gradient echosequence with 12 echo times (TR ) 100 ms, TE ) 3.1-44.8ms) was repeated 200 times over 50 min. The images wereanalyzed with OsiriX DICOM viewer, ImageJ (National Insti-tutes of Health, USA), and Prism 5 (GraphPad Software). Curve

fits for T2 calculation and enzyme inhibition (logarithmicinhibitor response curve, standard slope) were done withPrism 5.

Molecular Modeling. The 5 nm octahedral magnetite coreof VSOP-C200 with approximately 2600 iron atoms and 75citrate shell molecules per core (data provided by FerropharmGmbH) was generated with CrystalMaker (CrystalMaker Soft-ware). Molecular modeling of MMP-9-PSOP was done basedon the size measurements and the sequence of the MMP-9peptide and NHS-mPEG (MW 5000) using the PyMOL Mo-lecular Graphics System (12).

RESULTS

To add enzyme-sensing capability, we designed a constructshown in Figure 1a consisting of a peptide and a methyl-poly(ethylene glycol) polymer (mPEG, molecular weight 5000).The peptide consists of a cleavage domain with the enzymerecognition motif and a highly positively charged, arginine-richcoupling domain, which are connected by a linker sequence.For analysis by fluorescence methods, a fluorescine is coupledto the peptide. In a first step, this peptide was reacted with NHS-mPEG at the end of the cleavage domain. In a second step,after purification by gel filtration and HPLC control for purity,these peptide-mPEG copolymers were mixed in different ratioswith the VSOP. A ratio of 6 peptide-mPEG per VSOP yieldedMMP-9-PSOP with a hydrodynamic diameter of 24.9 ( 7.0nm (Figure 2b). Mixtures with ratios between 6 and 16 peptide-mPEG consistently yielded particles with sizes around 24 nm,

Figure 1. Synthesis and function of MMP-9-PSOP. (a) Synthesis. In step (1), the 25 amino acid peptides consisting of an arginine-rich couplingdomain and a cleavage domain with the recognition site for MMP-9 linked by a glycine bridge are reacted with NHS-methyl-poly(ethylene glycol)(NHS-mPEG). The resulting peptide-mPEG copolymers are purified by gel filtration to remove unreacted mPEG and NHS. In step (2), peptide-mPEG are mixed with VSOPs, yielding MMP-9-PSOP. (b) Function. When sterically stabilized PSOP with an intact mPEG shell (1) are exposedto a protease specific to the cleavage motif, the peptide-mPEG are cut at the cleavage domain, resulting in a loss of sterical stabilization (2). Theremaining particles aggregate due to the magnetic attraction of the iron oxide cores and the electrostatic attraction of the positive (arginine-richcoupling domain) and negative charged surface areas (acid shell).

Protease Nanosensors for MRI Bioconjugate Chem., Vol. 19, No. 12, 2008 2441

whereas ratios below 6 resulted in particles over 30 nm in size,probably due to insufficient steric stabilization (Figure 3 andTable 1).

To evaluate the level of copolymer coupling to the VSOPsurface, we measured the fluorescence quenching effect of thepeptide-bound fluorescein, which is caused by nonradiativeenergy transfer due to the proximity of the dyes to each otherand to the iron oxide cores (13). Compared with the fluorescenceof the peptide-mPEG complexes without the particles, thefluorescence was strongly quenched up to ratios of 8 complexesper VSOP (Figure 3). PSOP with 6 complexes per VSOP hada quenched relative fluorescence of 0.02, indicating nearly

complete coupling of the complexes to the VSOP surface.Higher ratios of complexes to particle cores showed a decreasingquenching effect, revealing incomplete coupling, most likelydue to saturation of the VSOP surface. When free fluoresceinwas mixed with the particles (6 fluorescein dyes per VSOP) ascontrol, the fluorescence was only reduced to 0.78, confirmingthat the reduction of copolymer fluorescence was not onlycaused by light absorption by the particles. Therefore, 6 peptide-mPEG copolymers per particle was considered a good ratio forfurther experiments.

For illustration, a model of MMP-9-PSOP (6 peptide-mPEGper particle), based on the sequence of the peptide-mPEGcomplexes and particle size measurements, is shown togetherwith a crystal structure of MMP-9 in Figure 2c.

The function of the particles is explained in Figure 1b. Whenthe sterically stabilized MMP-9-PSOP (1) are exposed to MMP-9, the protease cleaves the peptide at the recognition site (2),resulting in loss of the sterically stabilizing mPEG shell. Dueto the superparamagnetic properties of the iron oxide cores andthe ambivalent surface of the remaining particles with positively(coupling domains) and negatively (citrate coat of VSOP)charged areas, the particles aggregate driven by magnetic andelectrostatic attraction (3).

This process has two important consequences: First, clusteringof the superparamagnetic nanoparticles causes a substantialincrease in R2-relaxivity, called magnetic switch (6). Second,the particles are converted from mPEG-covered stealth particlesinto highly aggregative particles with strongly charged surfaces(14). Conveniently, mPEG-5000 has been shown to be optimalfor achieving stealth properties for nanoparticles (15). Conse-quently, once injected, the intact PSOP should remain for a longtime in the blood circulation until they reach an MMP-9expressing target tissue, where they are converted into aggre-gative particles and accumulate.

In contrast to mPEG-5000, peptide-mPEG copolymers madeof mPEG-2000 (molecular weight 2000) or of branched mPEG-2000 did not yield stable particles but precipitating aggregates,presumably due to insufficient steric stabilization (data not

Figure 2. Size and model of MMP-9-PSOP. The hydrodynamic size distribution of the parent VSOP with 7.7 ( 2.1 nm is shown in (a). With thecopolymers attached to the surface, the PSOP are still small at 25 nm with a narrow size distribution (b). For illustration, a model of MMP-9-PSOPbased on the peptide sequence, the structure of mPEG, and the size measurements is shown in (c). The parent VSOP consists of a 5 nm magnetitecore (gray) covered by a negatively charged citrate shell (red). The peptide-mPEG copolymers are electrostatically bound by the positively chargedcoupling domains (blue). The mPEG polymers (light blue) are linked to the coupling domain via the cleavage domain (yellow) and a linker peptide.A model of MMP-9 (22) (brown) is shown for size comparison at the cleavage domain. Fluorescein dyes (green).

Figure 3. Dependence of size and fluorescence quenching on the ratioof peptide-mPEG copolymers to VSOP. Left axis: With an increasingnumber of peptide-mPEG copolymers per VSOP, the hydrodynamicsize (mean of distribution by number, SD) of MMP-9-PSOP decreasesslightly up to a ratio of 6 peptide-mPEG per VSOP and stays constantaround 24 nm up to a ratio of 16. An average ratio of 6 peptide-mPEGper VSOP seems to be necessary for good steric stabilization. Belowthat ratio, the particles most likely form more dimers or multimers.Right axis: Due to the proximity of the copolymer-coupled fluoresceindyes to each other and to the nanoparticle cores, strong quenching offluorescence can be observed for PSOP with ratios up to 8, provingnearly complete binding of the copolymers to the VSOP. With higherratios, the quenching effect decreases, indicating incomplete couplingof copolymers due to saturation of the VSOP surface with peptide-mPEG copolymers.

2442 Bioconjugate Chem., Vol. 19, No. 12, 2008 Schellenberger et al.

shown). Another control experiment with uncoupled peptide andmPEG-5000 did not result in stable particles either (data notshown).

To demonstrate the function of MMP-9-PSOP, we measuredthe changes in size and T2* contrast in MRI following activationby MMP-9. Results are shown in Figure 4. First, we tested theactivation of MMP-9-PSOP by monitoring hydrodynamicdiameters (Figure 4a). Incubation of MMP-9-PSOP (6 peptide-mPEG per particle) with MMP-9 caused dramatic aggregationalready after a few minutes leading to particle sizes exceedingthe micrometer range. When the experiment was repeated inthe presence of a MMP-9 inhibitor, the size remained almostunchanged over a period of 2 h.

To monitor the activation using MRI, we prepared 200 µLreactions with different concentrations of MMP-9-PSOP (300nm, 150 nM, 75 nM, and 38 nM) and 1.3 U MMP-9 each.Approximately every 15 s, T2*-weighted images (gradient echosequences with 12 echo times (TE) between 3.1 and 44.8 msand a repetition time (TR) of 100 ms) were acquired over aperiod of 50 min. Figure 4b shows the time course of the signalintensities (for a TE of 10.6 ms). Depending on the concentra-

tion, the signal intensities decrease and reach a minimum atabout 15 min for a concentration of 75 nM concentration andat about 27 min for 300 nM. The signal decrease was about79% for 300 nM and 24% for 38 nM. The differences in thetime it took to reach the signal minimum can be explainedby the different amounts of peptide substrates to be cleaved bythe MMP-9. The MR image in Figure 4b represents a snapshottaken at 17.6 min. Figure 4c shows the time course of signalintensities measured under the same conditions for 300 µM-MMP-9-PSOP incubated with different concentrations of aMMP-9-inhibitor. The signal drop is delayed at a concentrationof 10 nM and nearly completely suppressed for the observationperiod of 50 min at a MMP-9-inhibitor concentration of 1 µM.T2* relaxation times for the different inhibitor concentrationswere calculated from the MR images obtained at 32 min andanalyzed using an inhibition function of Prism 5 software(log(inhibitor) vs response curve with standard slope, R2 )0.996). The 50% inhibitory concentration in this assay was 9.3nm (95% confidence interval 4.2 nM to 21 nM).

DISCUSSION

After mixing with the VSOP, the peptide-mPEG copolymersadsorb at the strongly negatively charged VSOP citrate surfacewith their positively charged coupling domains. The resultingprotease-specific iron oxide particles (PSOP) with the copoly-mers at the surface are no longer electrostatically but stericallystabilized, i.e., the natural tendency of superparamagneticmagnetite nanocores to aggregate is countered by the thickmPEG coat and not by the negatively charged surface of the

Figure 4. Function of PSOP. (a) Hydrodynamic size measurements. Time course of aggregation-induced size increase of the MMP-9-PSOP probe(0.125 nmol particles) induced by enzyme activation with 7.4 U of MMP-9. There was no probe activation when the experiments were done in thepresence of 30 µM MMP-9-inhibitor. (b-c) MRI experiments. (b) Time course of activation of MMP-9-PSOP at four different particle concentrations.As aggregates form, the T2*-relaxivity of these particles increases, causing a drop in signal intensities on the gradient echo sequences to a minimumuntil no further decrease occurs because continuing aggregation and precipitation reduce the amount of water protons affected. MR image obtainedat the time point indicated by the arrow. (c) Time course of signal intensities of MMP-9-PSOP (300 nM particle concentration) incubated withdifferent concentrations of a MMP-9-inhibitor. (d) Inhibition of the T2*-time-shortening at the time point indicated by the arrow in (c). The 50%inhibitory concentration in this assay is approximately 10 nM. MR experiments were done with 1.3 U MMP-9 in 200 µL buffer.

Table 1. Properties of MMP-9-PSOP

size [nm] 24.9 ( 7.0

Peptide-mPEG copolymers per particle 6R1 relaxivity [mM Fe1-s-1] 8.9R2 relaxivity [mM Fe1-s-1] 41Fe atoms per particle* 2600

* Data provided by Ferropharm GmbH. Relaxivities at 0.94 T.

Protease Nanosensors for MRI Bioconjugate Chem., Vol. 19, No. 12, 2008 2443

VSOP as before. Instead of PEG, other macromolecules suchas dextran could provide sterical stabilization as well.

The results of the size measurements and in Vitro MRIexperiments presented here confirm the specific protease-sensingfunction of these newly developed PSOP. Upon incubation withMMP-9, the nanosensors start clustering after a few minutes,as can be shown by dynamic laser light scattering and MRI.This activation by protease activity was largely suppressed bypreincubation with an MMP-2/MMP-9 inhibitor. Inhibitionoccurred in a typical, concentration-dependent manner (Figure4d), proving that activation was due to the protease activity ofMMP-9.

The PSOP comprise several key characteristics making thema good candidate for an enzyme reporter probe for in ViVo MRI:Synthesis of the particles is fairly straightforward. They can beadapted to any proteases or possibly other cleaving enzymes.Poly(ethylene glycol)s have been proven to be biocompatible(listed in the pharmacopeia) and provide stealth properties,which is a prerequisite for a sufficiently long circulation timein ViVo (15). Together with the nanosensors’ small diameter ofabout 25 nm, these properties ensure good bioavailability inthe desired target tissues. Unlike the caspase-3-sensitive nanosen-sors presented by Josephson and co-workers (6), which areclusters of approximately 45 nm sized CLIO monomers beforeactivation and are cleaved into monomers by caspase-3, leadingto a decrease in relaxivity and contrast effect, PSOP becomelarger upon activation and relaxivity increases. This propertyallows the in ViVo application of the nanoparticles. The parentVSOP has been shown to be safe in preclinical (16) and clinicaltrials (7). In contrast to near-infrared fluorescence imaging, MRIwith PSOP is a potential candidate for whole-body imaging ofprotease activity in humans. Moreover, the particles could beused to deliver therapeutics, or the accumulating magnetic coresthemselves could serve as “theranostics” in clinical magneticthermotherapy of cancer, which currently requires direct injec-tion of magnetic nanoparticles into the tumor (17).

The in Vitro experiments performed in this study have somelimitations: First, the size measurements yielded meaningfulresults only until the aggregates started to precipitate, therebydropping out of the measurement field of the zetasizer. In theMR experiments, the T2* contrast effect also started to decreaseat a certain point in time, which is partially attributable toprecipitation as well. Additionally, the fraction of water protonsinfluenced by the magnetic field of the particles also decreasesfrom a certain time onward as the aggregates continue toincrease in size while their number decreases (18). Althoughprecipitation due to aggregation is problematic for these in Vitrosettings, it is an intended and important mode of function forin ViVo application. The accumulation of the particles inprotease-expressing target tissues following the switch from aPEGylated stealth state to a highly surface-charged, aggregativestate cannot be demonstrated with the types of experiments usedhere and remains to be shown in ViVo. Although the aggregationeffect and consequently the magnetic switch would be desirablefor an enhanced contrast effect in the in ViVo situation, it is nota necessity, since a particle binding to the target tissue due tothe exposed charged surface is sufficient to produce a strongcontrast effect (19-21).

We are convinced that the nanosensors presented here havegreat potential as reporter probes for assessing enzyme activityof proteases by in ViVo MRI and can thus help diagnose diseaseswith pathologically altered protease activity or monitor proteaseinhibitor therapies.

ACKNOWLEDGMENT

VSOP was provided by Ferropharm GmbH. We thank BettinaHerwig for language editing.

LITERATURE CITED

(1) Weissleder, R., Tung, C. H., Mahmood, U., and Bogdanov,A. J. (1999) In vivo imaging of tumors with protease-acti-vated near-infrared fluorescent probes. Nat. Biotechnol. 17,375–378.

(2) Ntziachristos, V., Ripoll, J., Wang, L. V., and Weissleder, R.(2005) Looking and listening to light: the evolution of whole-body photonic imaging. Nat. Biotechnol. 23, 313–320.

(3) Weissleder, R., and Pittet, M. J. (2008) Imaging in the era ofmolecular oncology. Nature 452, 580–589.

(4) Louie, A. Y., Huber, M. M., Ahrens, E. T., Rothbacher, U.,Moats, R., Jacobs, R. E., Fraser, S. E., and Meade, T. J. (2000)In vivo visualization of gene expression using magnetic resonanceimaging. Nat. Biotechnol. 18, 321–325.

(5) Bogdanov, A., Matuszewski, L., Bremer, C., Petrovsky, A.,and Weissleder, R. (2002) Oligomerization of paramagneticsubstrates result in signal amplification and can be used for MRimaging of molecular targets. Mol. Imaging 1, 16–23.

(6) Perez, J. M., Josephson, L., O’Loughlin, T., Hogemann, D.,and Weissleder, R. (2002) Magnetic relaxation switchescapable of sensing molecular interactions. Nat. Biotechnol. 20,816–820.

(7) Taupitz, M., Wagner, S., Schnorr, J., Kravec, I., Pilgrimm, H.,Bergmann-Fritsch, H., and Hamm, B. (2004) Phase I clinicalevaluation of citrate-coated monocrystalline very small super-paramagnetic iron oxide particles as a new contrast medium formagnetic resonance imaging. InVest. Radiol. 39, 394–405.

(8) Chen, J., Tung, C. H., Allport, J. R., Chen, S., Weissleder, R.,and Huang, P. L. (2005) Near-infrared fluorescent imaging ofmatrix metalloproteinase activity after myocardial infarction.Circulation 111, 1800–1805.

(9) Kridel, S. J., Chen, E., Kotra, L. P., Howard, E. W., Mobashery,S., and Smith, J. W. (2001) Substrate hydrolysis by matrixmetalloproteinase-9. J. Biol. Chem. 276, 20572–20578.

(10) Hu, J., Van den Steen, P. E., Sang, Q. X., and Opdenakker,G. (2007) Matrix metalloproteinase inhibitors as therapy forinflammatory and vascular diseases. Nat. ReV. Drug DiscoVery6, 480–498.

(11) Bremer, C., Tung, C. H., and Weissleder, R. (2001) In vivomolecular target assessment of matrix metalloproteinase inhibi-tion. Nat. Med. 7, 743–748.

(12) DeLano, W. L. (2002) The PyMOL Molecular GraphicsSystem, DeLano Scientific, USA.

(13) Josephson, L., Kircher, M. F., Mahmood, U., Tang, Y., andWeissleder, R. (2002) Near-infrared fluorescent nanoparticles ascombined MR/optical imaging probes. Bioconjugate Chem. 13,554–60.

(14) Moghimi, S. M., and Szebeni, J. (2003) Stealth liposomes andlong circulating nanoparticles: critical issues in pharmacokinetics,opsonization and protein-binding properties. Prog. Lipid Res. 42,463–478.

(15) Shi, B., Fang, C., and Pei, Y. (2006) Stealth PEG-PHDCAniosomes: effects of chain length of PEG and particle size onniosomes surface properties, in vitro drug release, phagocyticuptake, in vivo pharmacokinetics and antitumor activity.J. Pharm. Sci. 95, 1873–1887.

(16) Wagner, S., Schnorr, J., Pilgrimm, H., Hamm, B., and Taupitz,M. (2002) Monomer-coated very small superparamagnetic ironoxide particles as contrast medium for magnetic resonanceimaging: preclinical in vivo characterization. InVest. Radiol. 37,167–177.

(17) Thiesen, B., and Jordan, A. (2008) Clinical applications ofmagnetic nanoparticles for hyperthermia. Int. J. Hyperthermia1–8.

(18) Koh, I., Hong, R., Weissleder, R., and Josephson, L. (2008)Sensitive NMR sensors detect antibodies to influenza. Angew.Chem., Int. Ed. Engl. 47, 4119–4121.

(19) Kaufels, N., Korn, R., Wagner, S., Schink, T., Hamm, B.,Taupitz, M., and Schnorr, J. (2008) Magnetic resonance imaging

2444 Bioconjugate Chem., Vol. 19, No. 12, 2008 Schellenberger et al.

of liver metastases: experimental comparison of anionic andconventional superparamagnetic iron oxide particles with ahepatobiliary contrast medium during dynamic and uptakephases. InVest. Radiol. 43, 496–503.

(20) Schnorr, J., Wagner, S., Abramjuk, C., Drees, R., Schink, T.,Schellenberger, E. A., Pilgrimm, H., Hamm, B., and Taupitz,M. (2006) Focal liver lesions: SPIO-, gadolinium-, and ferucar-botran-enhanced dynamic T1-weighted and delayed T2-weightedMR imaging in rabbits. Radiology.

(21) Schellenberger, E., Schnorr, J., Reutelingsperger, C., Un-gethum, L., Meyer, W., Taupitz, M., and Hamm, B. (2008)Linking proteins with anionic nanoparticles via protamine:ultrasmall protein-coupled probes for magnetic resonance imag-ing of apoptosis. Small 4, 225–230.

(22) Browner, M. F., Smith, W. W., and Castelhano, A. L. (1995)Matrilysin-inhibitor complexes: common themes among metal-loproteases. Biochemistry 34, 6602–6610.

BC800330K

Protease Nanosensors for MRI Bioconjugate Chem., Vol. 19, No. 12, 2008 2445