Disruptive chemical doping in a ferritin-based iron oxide nanoparticle to decrease r₂ and enhance detection with T₁-weighted MRI

Abstract

This work reports chemical doping to create flexible, paramagnetic nanoparticle contrast agents for in vivo molecular magnetic resonance imaging (MRI) with low transverse relaxivity (r₂). Most nanoparticle contrast agents formed from superparamagnetic metal oxides are currently developed with high r₂. While sensitive, they can have limited in vivo detection due to a number of constraints with T₂ or T₂*-weighted imaging. T₁-weighted imaging is often preferred for molecular MRI, but most T₁-shortening agents are small chelates with low metal payload or are nanoparticles that also shorten T₂ and limit the range of concentrations detectable with T₁-weighting. Here we used tungsten and iron deposition to form doped iron oxide crystals inside the apoferritin cavity to form a WFe nanoparticle with a disordered crystal and un-coupled atomic magnetic moments. The atomic magnetic moments were thus localized, resulting in a principally paramagnetic nanoparticle. The WFe nanoparticles had no coercivity or saturation magnetization at 5K and sweeping up to +/−20000Oe, while native ferritin had a coercivity of 3000Oe and saturation at +/−20000Oe. This tungsten-iron crystal paramagnetism resulted in an increased WFe particle longitudinal relaxivity (r₁) of 4870mM⁻¹s⁻¹ and a reduced transverse relaxivity (r₂) of 9076mM⁻¹s⁻¹ compared to native ferritin. The accumulation of the particles was detected with T₁-weighted MRI in concentrations from 20nM-400nM in vivo, both injected in the rat brain and targeted to the rat kidney glomerulus. The WFe Apoferritin nanoparticles were not cytotoxic up to 700nM particle concentrations, making them potentially important for targeted molecular MRI.

1. INTRODUCTION

Sensitivity is an important challange for in vivo molecular magnetic resonance imaging (MRI). There is great interest in developing contrast agents to detect cells and molecules in vivo. Most commercial agents are detected in μM-mM concentrations, yet many biological molecular targets are present in pM-nM concentrations (1,2). There is thus a strong need for contrast agents for magnetic resonance imaging (MRI) that are detectable in sub-μM concentrations.

Researchers often improve MRI contrast agents by increasing the transverse (T₂ or T₂*) or longitudinal (T₁) relaxivity. The relaxivity (r₂ or r₁) is a model of the impact of the agent on the relaxation times and consequently on image contrast. Since most tissues have longer intrinsic T₁ compared to T₂ (3), a T₁-shortening agent can be detected at ~30-fold lower concentrations than a T₂-shortening agent of similar relaxivity. T₁-shortening agents may also be more readily distinguished in practice because T₂ or T₂* changes can resemble intrinsic image artifacts. T₁-shortening agents are often formed by the localized magnetic moment of unpaired electrons in a valence orbital of single atoms. In comparison, many T₂- and T₂*-shortening contrast agents are often formed by exchange coupling between multiple atoms in an oxide crystal. This coupling yields a greater net magnetic moment than the sum of the uncoupled moments of the component metal atoms. Thus, while T₁-shortening agents are often more desirable, T₂- and T₂* shortening agents can be more readily conferred a high relaxivity and metal payload. An example of this phenomenon is seen in the development of magnetic nanoparticles. Magnetic nanoparticles are useful as contrast agents because they have a high metal payload, are often stable in vivo, and can be readily functionalized. Unfortunately, attempts to create nanoparticle agents with high r₁ usually result in a simultaneous enhancement in r₂ and r₂*(4–8). This limits the range of concentrations over which the agent can be detected with T₁-weighted imaging and limits the sensitivity to the imaging target.

Several groups have developed T₁-shortening agents with enhanced r₁ (4,6,8–13). Kim et al reported a metal oxide based 2–10 nm particle with a relatively low r₂/r₁ of ~4–25 (6). The smallest nanoparticles had a low r₂ because of surface defects that created localized, rather than de-localized spins. Modulating r₂/r₁ with nanoparticle size is an important advance in nano-scale magnetism and magnetic resonance relaxation.

To take advantage of the high relaxivities of superparamagnetic iron oxides, several pulse sequences have been developed to create a bright signal at the location of a T₂-shortening agent. Techniques such as ultrashort echo time (UTE), off-resonance saturation, and “bright blood” imaging have been used to obtain T₁ weighting regardless of the induced short T₂ and T₂*. These techniques are successful, but are also limited by high RF deposition, difficulty of multislice 3D imaging, and cumbersome superposition of anatomical and target imaging(14,15). Positive contrast has been successfully applied for in vivo imaging (16–20), but detection is still limited to a concentration window where r₂ and r₂* induced field artifacts do not overwhelm the T₁ shortening(21). Having the ability to tune r₂/r₁ for even higher sensitivity of detection could circumvent this problem and offer a wider range of concentrations for imaging. Hence, tuning r₂/r₁ in a specific particle could offer the versatility needed to improve imaging with either conventional T₁-weighting sequences, or make sophisticated sequences such as UTE imaging even more powerful. There is thus a need for a general, flexible technique to create nanoparticles with low r₂/r₁.

To address this need, we propose chemical doping to create a T₁-shortening nanoparticle with low transverse relaxivity. In crystals of 3d transition metals, the atoms are often coupled through electron exchange. Paramagnetic metals contain atoms with localized magnetic moments, either because the electronic wave functions of the atoms do not overlap, or because they are physically separated, as illustrated in Figures 1a and 1b. If a dopant is strategically incorporated into the metal crystal lattice, it may be possible to disrupt exchange coupling and localize the magnetic moments of the component atoms, creating a paramagnetic, T₁-shortening nanoparticle contrast agent with a high metal payload. If established, chemical doping could facilitate metal oxide nanoparticles with tunable size and r₂/r₁.

Figure 1.

Schematic of atomic spin coupling in a nanoparticle to form a superparamagnetic or paramagnetic crystal. (a) De-localized electron spins in a super-paramagnet and localized electron spins in a paramagnet, both exposed to an external magnetic field. The size of the magnetic moments (arrows) in the superparamagnetic crystal compared with those of the paramagnetic crystal illustrate the difference in magnetization strength owing to the weaker uncoupled magnetic moments in the paramagnetic crystal. (b) Metal and oxygen atoms in a spinel arrangement with octahedral and tetrahedral sites occupied by Fe2+ or Fe3+ atoms. 3d orbital and p orbital overlap, facilitating spin coupling. (c) Randomly aligned moments without an external field and aligned moments with an external magnetic field inside the apoferritin shell. Water molecules in close proximity interact with the surface of the paramagnetic crystal core.

To investigate this concept, we formed a mixed metal oxide in the apoferritin protein. Apoferritin is composed of 24 subunits (of either H- or L- forms) that form a spherical cage with ~ 13 nm diameter and a ~8nm core. The H-ferritin subunit contains a ferroxidase that converts Fe²⁺ to Fe³⁺. The ferroxidase can be used to create a nanoparticle contrast agent from apoferritin, suitable for in vivo targeting and delivery for MRI(11,22–28). A metal/metal oxide-loaded apoferritin is useful as a “natural” contrast agent for MRI because it is uniform in size, contains multiple surface functional groups, and is relatively immunologically inert (29,30).

2. RESULTS AND DISCUSSION

2.1. Synthesis and Characterization of Tungsten Iron Apoferritin Nanoparticle

To create the WFe composite, we first seeded an iron oxide core in apoferritin, as previously described (25), using the intrinsic ferritin ferroxidase. We then incorporated both tungsten and iron into the apoferritin core. Tungsten was incorporated at different rates. We selected the configuration that yielded the highest relaxivity (See the Supporting Information section for details regarding synthesis with varying tungsten additions). After synthesis, dialysis, and purification, the ferritin nanoparticle core comprised an electron-dense metal crystal about 6nm in diameter, as demonstrated with transmission electron microscopy (TEM, Figure 2a). The WFe-apoferritin yield was 10%. Tungsten has been previously used as a dopant to create superconductive composites and as gas sensors(31–33). Tungsten-doped iron oxides have been used to form porous and semi-amorphous wolframites with large regions of uncoupled spins (34). The tungsten ions can block coupling by superexchange and do not readily accept electrons from the oxygen bridges. Tungsten electrons should therefore not exchange with neighboring iron atoms, as illustrated in Figure 1b. If formed inside apoferritin, a tungsten-doped metal crystal could disrupt the spin exchange and magnetic order typically found in these materials, forming an amorphous crystal core (Figure 1c). An amorphous crystal would have distinct advantages as a contrast agent because it has reduced r₂/r₁ and provides a large surface area for water proton exchange.

Figure 2.

Electron microscopy confirmed the formation and composition of an amorphous WFe crystal inside apoferritin. (a) Transmission electron microscopy of WFe composite cores showing ~6–8 nm cores. (b) High-resolution electron microscopy of a WFe nanoparticle showing no structured crystal. Fast Fourier transform pattern in the inset shows no crystalline diffraction spots. (c) Energy dispersive spectrum of a selected region of WFe apoferritin. Peaks correspond to iron and tungsten transitions. Carbon and copper peaks are from the grid. Insets: scanning transmission electron microscopy showing particle selection for line profile plot. Line profile indicates a ~6–8 nm core with both W and Fe. Curves were offset to avoid overlap. Amplitude does not indicate metal concentration. (d) X-Ray diffraction peaks showing the polymorphic crystal characteristic of of native ferritin (top spectrum), and WFe nanoparticle spectrum showing an amorphous crystal (bottom spectrum). Selected area electron diffraction confirmed the WFe nanoparticle lack of crystallinity showing only a diffuse ring arising from the selected area (bottom panel). Native ferritin had several diffraction rings with d-spacings characteristic of magnetite/maghemite.

The metal cores of the WFe nanoparticles were homogeneously filled (Figure 2a). There were no visible lattice fringes in high-resolution electron microscopy (HREM, Figure 2b). The Fourier transform of the HREM image of the particle, shown in the inset of Figure 2b, contained no diffraction spots, indicating the absence of a crystalline lattice. The WFe-apoferritin nanoparticles were visible in scanning transmission electron microscopy (STEM), as shown in Figure 2c. Based on the energy dispersive spectrum (EDS) across the particle, the core diameter was ~7 nm, consistent with the inner diameter of apoferritin. Line profiles of the EDS signal, shown in Figure 2c, indicated that tungsten and iron were both evenly distributed in the core, suggesting that both metals were incorporated into the crystal lattice. WFe nanoparticles exhibited no X-ray diffraction patterns, (Figure 2d, bottom panel), confirming that cores of the WFe nanoparticles were amorphous. In contrast, the cores of native ferritin exhibited the characteristic peaks of magnetite/maghemite, (Figure 2d, top panel). WFe nanoparticles exhibited no distinct d-spacings in selected area electron diffraction (SAED), while native ferritin exhibited the characteristic d-spacings for magnetite/maghemite of 0.25nm, 0.205nm, 0.154nm, and 0.130nm (Figure 2d, top panel). There were approximately 8788 ± 3272 metal atoms (8743±3277 iron and 77±55 tungsten) inside each nanoparticle, measured by inductively-coupled plasma-optical emission spectroscopy (ICP-OES). This confirmed that an amorphous tungsten-doped nanoparticle had formed inside the apoferritin protein.

We characterized the distribution of the WFe-apoferritin particle sizes, monodispersity, and stability with size exclusion chromatography (SEC) and native gel electrophoresis, as detailed in Figure 3. The WFe-apoferritin particles were monodisperse and stable after weeks of storage, making them practical for in vivo imaging experiments.

Figure 3.

(a) Native tris-glycine gel electrophoresis showing that similar molecular weights of the Wfe composite, native ferritin and apoferritin. (b) Photograph showing concentrated WFe particle solution (left) and solution after purification (right). (c) Size exclusion chromatography of WFe nanocomposite, native ferritin and apoferritin, showing same elution time of protein (280 nm) before and after mineralization. Metal (410 nm) in the WFe and native ferritin eluted at the same time and was not present in apoferritin.

2.2. Magnetic Properties of Tungsten-Iron Apoferritin Nanoparticles

To determine whether tungsten doping made the nanoparticle paramagnetic, we performed magnetometry using a superconducting quantum interference device (SQUID). Figure 4a shows the SQUID magnetization curves at 5K. The WFe nanoparticles exhibited increasing magnetization with applied field (up to 3T), almost no saturation, and showed no hysteresis, all characteristics of paramagnetism. In contrast, native ferritin exhibited saturation at ±20000Oe and hysteresis, characteristic of superparamagnetism. The effective magnetic moment of the WFe particle was ~63μ_B based on temperature-varied measurements of magnetization reversal (Figure 4b and d). This moment is lower than the reported magnetic moment for reconstituted ferritin of 2590μ_B (35). Field cooled (FC) and zero field cooled (ZFC) magnetization measurements as seen in Figure 4c indicated that the paramagnetic WFe nanoparticles had a reduced blocking temperature (~3K) compared to that of native ferritin (~12K). The Curie-Weiss behavior of the magnetization curves shown in Figure 4d indicated an exchange interaction between magnetic moments of atoms in the crystal in addition to the interaction with the external magnetic field. The positive temperature intercept (~9K) suggests that the inter-atomic interaction is ferromagnetic. Taken together, the magnetization measurements confirmed that the WFe-apoferritin nanoparticles were paramagnetic.

Figure 4.

Superconducting quantum interference device (SQUID) magnetometry for WFe-apoferritin nanoparticles. (a) SQUID reversal curves for WFe nanoparticles and native ferritin at 5 K showing paramagnetism in WFe nanoparticles and superparamagnetism in native ferritin. (b) Temperature-dependent magnetization reversals were performed at 5, 10, 50 and 150 K. (c) The slope of the curve as a function of temperature was measured to calculate the magnetic susceptibility and the Curie constant. The WFe nanoparticles were paramagnetic at different temperatures and based on the line fit to the susceptibility vs temperature (d) The WFe nanoparticles exhibited Curie-Weiss paramagnetic behavior above a temperature of ~9 K and ferromagnetic behavior below that temperature.

2.3. Relaxometry Measurements

The observed paramagnetism of the WFe nanoparticles should make them T₁-shortening MRI contrast agents. To confirm this and to estimate the range of necessary MRI pulse sequence parameters for detection, we measured the per-metal and per-particle r₁ and r₂. The particle relaxivity was measured from the protein concentration, (since the apoferritin particle is always formed from the same amount of protein), and the metal relaxivity was measured from the total metal concentration. The WFe nanoparticles had a particle r₁ of 4870 ± 1199mM⁻¹s⁻¹, total metal r₁ of 0.63 mM⁻¹s⁻¹ and an r₂/r₁ ratio of 1.86 at a magnetic field strength of 1.5T, as summarized in Table 1. Although our total atomic relaxivity is lower than that of some conventional paramagnetic ionic contrast agents, it is critical to note that the net atomic relaxivity is essentially irrelevant for T₁-shortening nanoparticle contrast agents in determining the concentrations of targeted molecules that can be detected in vivo. Atoms inside the core of the crystal in a particle of this size do not typically interact with the water protons to change T₁. Since the dipolar interaction between the magnetic moment of the water protons and the electronic moment of the paramagnetic metal is local and effective only at short distances (e.g. 2–3Å), the increased r₁ of the WFe nanoparticle is likely due to an interaction with protons at the outer surface of the crystal (refer to Supporting Information for more details). In computing the atomic relaxivity, the paramagnetic moments from metal atoms not in close contact with the surrounding water molecules (>1 atom from the surface) are therefore not to be included. Thus, particle relaxivity is a more accurate reflection of the detectable target concentration. For most nanoparticles the particle relaxivity is difficult to measure because the concentration of the particles can only be estimated. With ferritin, the nanoparticle concentration and particle relaxivity are readily determined from the protein concentration.

Table 1.

Relaxometry of WFe-apoferritin nanoparticles in vitro and comparison with similar paramagnetic agents.

	WFe-apoferritin based on particle concentration	WFe-apoferritin based on metal concentration	WFe – apoferritin based on surface metal atoms	Native Ferritin based on particle concentration	Mn-Apoferritin Nanocomposite based on particle concentration (8)	Gd2O3 based on metal concentration (13)	Gd-DTPA (42)
r1 (mM−1s−1)	4870.27	0.63	2.62	39	6650	9.9	3.75
r2 (mM−1s−1)	9076.67	1.04	4.89	3079	47200	10.5	4.89
r2/r1	1.86	1.67	1.86	78	7.09	1.1	1.30
Detectable Concentration Range with T1-weighted Contrast	21 – 400nM			N/A	15 – 77nM	N/A	27–741 μ M

All relaxivity measurements were performed at 1.5T, 37°C with particles suspended in 1% agarose. T₂ and T₁ relaxation times were obtained with a Carr-Prucell-Meiboom-Gill and an Inversion Recovery pulse sequence, respectively. Detectable concentrations were calculated assuming a 10% change in T₁ and 20% change in T₂ at the location of agent, and a background brain tissue T₁ and T₂ of 1084ms and 69ms (3), respectively. The concentration range is determined by the minimum and maximum concentration for detection with T₁-weighting before it can be detected with T₂-weighting. Previously reported apoferritin Mn-loaded nanoparticles have a smaller range of concentrations to detect with T₁-weighting (8). Gd₂O₃ nanoparticles and Gd-DTPA chelate are included here merely as a comparison to the WFe nanoparticles. Detectable concentrations of Gd₂O₃ nanoparticles were not included as the relaxivities for this computation were atomic relaxivities. Detectable concentration ranges were not included for Native ferritin as the minimum concentration to detect with T₁-weighted contrast superseded the minimum concentration at which T₂ effects would null the signal.

The increased particle r₁ in WFe nanoparticles may also be due to the enhanced proton exchange at the surface of apoferritin observed by Vasalatyi et al. (36). While outside the scope of this work, a complete investigation of water exchange through apoferritin via nuclear magnetic relaxation dispersion (NMRD) may be important to optimizing apoferritin relaxivity. To account for the expected strong surface contribution to r₁, we computed a “surface relaxivity” that is the atomic relaxivity of the estimated number of atoms on the crystal surface (Table 1). With a longitudinal particle relaxivity of 4870.27mM⁻¹s⁻¹ and r₂/r₁ of 1.86, WFe-apoferritin should be detectable with T₁-weighting in the brain, for example, in about 20nM-400nM concentrations, assuming a normal tissue T₁ and T₂ of 1084ms and 69ms, respectively(3). Mn-loaded apoferritin nanoparticles were previously introduced as sensitive T₂-shortening agents despite their relatively high particle r₁(8). This was likely because, based on our calculations, the high r₂ would yield particles with a narrow concentration detection range of 15nM-77nM with T₁-weighting. Above this range, the Mn-loaded nanoparticles are primarily detected as T₂-shortening agents. Similarly, native ferritin would not be readily detected as a T₁-shortening agent because the minimum concentration for detection with T₁-weighting is ~2.6μM, and the minimum for T₂-weighting is ~1.1μM. In comparison, particles with low r₂/r₁ such as Gd₂O₃ could have a large range of detectable concentrations. However, since the concentration of these particles is not readily measured it is not possible to infer sensitivity. The WFe nanoparticles are therefore advantageous for T₁-weighted MRI in a wide range of sub-μM concentrations, in vivo.

2.4. In vivo MRI of WFe-apoferritin Nanoparticles

To confirm that the WFe nanoparticles are detectable in vivo with T₁-weighted MRI, we injected 8ul of 195nM nanoparticles into the caudate/putamen of adult Sprague Dawley rat brains and the same volume of native ferritin in the contralateral side as a control. Figure 5b shows a typical in vivo T₁-weighted gradient-echo MRI (TE/TR=3.8ms/55.9ms) of the brain after injection, acquired at 7T. The injection site of the WFe nanoparticles was hyperintense (left side of the brain). Native ferritin was not detected (right side of the brain). The T₁ was 22% shorter at the site of WFe injection (2.02 s) than at the contralateral site of native ferritin (2.61 s), as shown in Figure 5c. See supporting information for in vivo imaging using a Multi Slice Multi Echo T₁-weighted sequence.

Figure 5.

WFe nanoparticles are detected in vivo with T1-weighted MRI. (a) 2D gradient echo T1-weighted image of a 1% agarose phantom. Increasing concentrations of WFe-apoferritin result in increased positive image contrast. (b) 3D Gradient echo T1-weighted image of an in vivo inoculation of 8 μl of 195 nm WFe nanoparticles and native ferritin into a rat striatum. Images were acquired on a 7 T Bruker small animal scanner. (c) T1 map color overlay showing an average T1 for each ROI at the injection sites. The site of the injection of WFe nanoparticle had an average T1 22% shorter than the contra lateral region of interest. (d) In vivo intravenous injection of paraCF labeling kidney glomeruli, spleen, and liver. (e) Control image of a rat in vivo with no injection of agent. (f) Thresholded image of paraCF labeled and (g) no injection of agent. Highlighted in red are pixels with intensities 50% higher than background muscle tissue, insets show cortex of paraCF kidney and control kidney. It is clear that the kidney is labeled with paraCF showing evident hyperintensities, while control did not show significant enhancement over background. (h) Immunofluorescence histology of a paraCF labeled kidney glomerular basement membrane and (i) native ferritin control. The kidney sections were labeled with DAPI (blue) and anti-horse spleen ferritin antibody with a rabbit anti-ferritin antibodyfollowed by a goat anti-rabbit IgG secondary antibody labeled with Alexa594 (red). Ferritin immunofluorescence (red) is present only in the paraCF labeled kidneys and not in the control kidneys. Scale bars 20 μm.

Although we did not experimentally compare the WFe nanoparticles to other contrast agents in varying doses, the r₁ at 7T of WFe nanoparticles of 2200mM⁻¹s⁻¹ (Figure 5a), the known background tissue T₁, and the measured shortened T₁ indicates that ~50nM was detected in vivo. Thus, WFe nanoparticles are useful as passive (un-targeted) agents with T₁-weighted MRI at concentrations in 10s of nanomolar at 7T.

To confirm that WFe-apoferritin nanoparticles could also be used as targeted contrast agents, we functionalized them with surface amine groups to create a paramagnetic form of cationized ferritin (“paraCF”). Cationized ferritin (CF) is an example of a targeted superparamagnetic contrast agent used to detect and count individual glomeruli in the kidney with MRI (37,38). Based on our experience, CF relaxivity and charge are maintained for at least several weeks to months when the CF is stored in solution at 4°C. CF also readily binds to the extracellular matrix of other organs containing fenestrated endothelia. However, CF is superparamagnetic and is therefore difficult to detect against the blood background in vivo (39). Also, paraCF may be more readily detected in nontoxic doses. To test the improved detection of labeled glomeruli in vivo with paraCF, we cationized WFe-apoferritin using published methods and injected it intravenously in rats (40).

There was significant enhancement in in vivo MRI (Figure 5d) in the kidney and spleen compared to the naïve controls (Figure 5e). We normalized the image to the magnitude surrounding muscle and set a threshold of 50% to map WFe-enhancement (Figure 5f). The labeled kidney cortex had punctate hyperintense spots, consistent in distribution with recent ex vivo results using CF (38). Images of control kidneys without paraCF labeling were clear of hyperintense spots (Figure 5g) and the presence of the paraCF particle in the kidneys was confirmed with immunohistochemistry. Figure 5h–i shows kidneys labeled with paraCF and no agent (Figure 5i). CF accumulates by electrostatic binding to proteoglycans in the glomerular basement membrane (37). ParaCF accumulated in the anionic basement membrane of kidney glomeruli, but native ferritin did not accumulate and was not visible with immunohistochemistry. We conclude that the paraCF nanoparticles are practical as targeted contrast agents in vivo.

2.5. W-Fe Apoferritin Nanoparticle Toxicity

Toxicity is important to the utility of any novel agent. CF has been shown to be neither nephrotoxic nor hepatotoxic when injected intravenously in MRI-detectable concentrations. CF was previously observed in the kidney, liver, lungs and spleen after intravenous injection (30). To investigate the cytotoxicity of the WFe-apoferritin, we incubated adherent 3T3 fibroblasts with 0 to 700nM of WFe apoferritin for 24 hours. Cells were ~97% viable after exposure to nanoparticle concentrations lower than 700nM, equivalent to ~5mM metal as seen in Figure 6. Therefore, our WFe nanoparticles are relatively non-toxic to cells in particle concentrations ranging from 3–700nM.

Figure 6.

WFe nanoparticles are nontoxic in MRI-detectable concentrations. (a) Fluorescence image of 3T3 fibroblasts incubated with WFe for 24 h. Live cells are labeled with calcein (green) and dead cells are labeled with EthD-III (red). (b) Viability and cytotoxicity quantitative results indicating no significant cell toxicity for particle concentrations lower than 700 nm. Error bars indicate the standard error of the mean (n=3).

3. CONCLUSION

We used chemical doping to create an amorphous WFe nanoparticle with high T₁ relaxivity. To our knowledge this is the first report of chemical doping to decrease r₂ and enhance detection of a T₁-shortening MRI contrast agent. Doping with tungsten, (in a ratio of 0.01to iron concentration in the crystal), disrupted the exchange coupling between spins to reduce r₂ and consequently reduce r₂/r₁. We detected the WFe-apoferritin nanoparticles in the brain in vivo in ~10s-100s of nM concentrations with T₁-weighted MRI. The WFe nanoparticles were also readily functionalized and detected in the glomerular basement membrane and spleen in vivo. We conclude that chemical doping can be used to create nanoparticles with low r₂/r₁ that are detectable in vivo with T₁-weighted MRI in nM concentrations.

4. MATERIALS AND METHODS

4.1. Particle Synthesis

To synthesize a tungsten and iron oxide filled apoferritin, 2μM apoferritin (Sigma Aldrich, St. Louis, MO) was buffered in 0.05M 2-(N-Morpholino) ethanesulfonic acid (MES) at pH 8.5. In separate containers we used 48mM FeCl₂ (Sigma Aldrich), and 48mM Na₂WO₄ (Sigma Aldrich). All solutions were de-aerated for at least 15 minutes with N₂, then kept in a water bath at 55 to 60°C under vacuum. Any apoferritin solution that evaporated during the de-aerating period was compensated by adding a previously de-aerated 0.05M MES buffer to maintain a constant volume throughout the reaction. We added the de-aerated 48mM FeCl₂ at a rate of 12.5μl/min to the apoferritin solution using a syringe pump for a total of 140 minutes. Fifty minutes into FeCl₂ delivery, the de-aerated 48mM Na₂WO₄ was added at a rate of 12.5μl/min using a syringe pump for a total of 40 minutes. A total of 1.75ml of FeCl₂ and 500μl of Na₂WO₄ were added to the protein solution. 200μl of 300mM sodium citrate was added to the solution to chelate any remaining metal ions. The solution was then sonicated for 10 minutes and spun for 10 minutes at 957 · g. Finally, the supernatant solution was collected and dialyzed overnight against 3L of de-ionized water using a 8,000 MW cut off dialysis tubing (BioDesignDialysis Tubing, Carmel, NY). Once dialyzed, the protein solution was filtered using 0.8μm and 0.2μm surfactant free cellulose acetate syringe filters (Corning Incorporated, Corning, NY) to rid the solution of coarse non-specific metal oxides. Total protein concentration was obtained with a Coomassie Plus Bradford Assay Kit (Thermo Scientific, Rockford, IL). Inductively coupled plasma – optical emission spectroscopy (ICP-OES) of particles suspended in 2% nitric acid was used to measure metal concentrations. We repeated WFe-apoferritin nanoparticle synthesis and analysis six times, with similar results.

4.2. ParaCF synthesis

Ferritin is readily cationized using a carbodiimide method outlined in Danon et al.(40) We use the same method to cationize the surface of the WFe ferritin nanoparticles, rendering what we refer to as paraCF. Briefly, approximately 0.5ml of 55mg/ml WFe-apoferritin nanoparticle was added to a 2ml solution of 2N N,N-dimethyl-11,3-propanediamine at pH 6. pH was adjusted with 2N NaOH and 0.2N HCl. Once pH was stable we added 200mg of 1-ethyl-3(3-dimethyl-aminopropyl) carbodiimide hydrochloride and monitored the reaction for 2h to maintain the pH at 6. The number of positive charges on the nanoparticle surface has been characterized and reported (40). Based on this protocol, approximately 60% of available surface carboxyls are activated and result in a relative net charge per molecule of approximately +244, compared to a net charge of −92 for native ferritin (40). The resulting solution was sealed and allowed to react overnight at room temperature. Once cationized, the resultant paraCF was dialyzed two times against phosphate buffered saline (PBS) and a single bolus injection of 11 mg paraCF in 1 ml PBS was administered intravenously, as described in Section 4.4.

4.3. Relaxometry

To mimic tissue intrinsic T₁ and T₂ at 1.5T, 500μl samples of different ferritin concentrations were prepared and mixed with 500μl of 2% agarose gel to yield a 1% agarose/sample mix (41). Each gel solution was refrigerated for at least 15 minutes and later brought up to 37°C before measurement. Relaxivity was measured using a MQ60 1.5T Bruker Minispec relaxometer (Bruker Optics, The Woodlands, TX). Bruker’s curve-fitting tool was used to find the corresponding T₂ values using a Carr-Purcell-Meiboom-Gill pulse sequence (CPMG) (Inter pulse τ = 4ms, TR = 15s, 75 points). T₁ values were obtained using the curve-fitting tool and the relaxation times were obtained with an Inversion Recovery pulse sequence (First TI = 5ms, last TI = 20,000ms, TR = 15s, 8 averages, 10 points). Samples were maintained at 37°C during measurements. Surface relaxivity was calculated by obtaining the number of iron atoms on the crystal surface. Refer to supporting information for the formulas used to calculate this metric. Relaxometry measurements at 7T were obtained from a T₁ map of a 1% agarose phantom at different concentrations and with a T₁-map RARE pulse sequence, TE/TR = 10.88ms/233.3ms, 500ms, 1200ms, 2500ms, 5000ms, NEX 1 and heated to ~37°C using a water-heated blanket.

4.4. In vivo imaging

All animal experiments were approved by the Arizona State University Institutional Animal Care and Use committee. Three adult male Sprague Dawley rats were anesthetized with a Ketamine - Xylazine cocktail (90mg/kg and 5mg/kg, respectively) and secured to a stereotactic frame (David Kopf Instruments, Tujunga, CA). The particles were injected into the striatum of a rat in the caudate/putamen, 1.6mm anterior and 3mm lateral from the bregma. To breach the skull, we created two burr holes in the skull bilaterally at the stereotactic coordinates using a drill. A 10μL Hamilton syringe needle (Sigma Aldrich, St. Louis) was inserted 6mm below the dura. The needle remained in place for 1 minute and then was retracted to 5mm below the dura. 8μL of WFe and control (native ferritin) were injected over the span of 5 minutes into the left and right hemispheres, respectively. The rat was imaged on a Bruker 7T/30 scanner (three-axis gradient, slew rate of 150 T/m/s) using a rat surface RF coil. 2% Isofluorane gas and oxygen-enhanced air were delivered during the scan. To image the striatum, a fast low-angle shot pulse sequence (FLASH) was used (TE/TR= 3.8ms/55.9ms, NEX 4), and a T₁-map RARE pulse sequence, TE/TR = 10.88ms/233.3ms, 500ms, 1200ms, 2500ms, 5000ms, NEX 1.

Two paraCF-injected rats were imaged 1.5 hr after injection using a custom made surface coil and a T₁-weighted 2D gradient recalled echo (GRE) MRI pulse sequence (TE/TR = 10/54, field of view = 3×3 cm, matrix size = 256×256, slice thickness = 400 um). Two naïve, un-injected rats were also imaged using the same MRI pulse sequence and parameters. Refer to Supporting Information for surface coil design.

ParaCF labeling of the kidney glomerulus was quantified by normalizing the signal magnitude of the renal cortex to the muscle surrounding the spine, (which was assumed to remain unlabeled by paraCF). The same normalization process was performed in the control rat.

4.5. Electron Microscopy

Samples were adsorbed on 400 mesh Cu-C grids (Provided by the WM Keck Bioimaging Laboratory) and transmission electron microscopy (TEM) images were obtained using a Philips CM12S Scanning Transmission Electron Microscope with EDAX 9800 plus an energy dispersive x-ray spectrometer fitted with a Gatan model 791 CCD camera for image acquisition (Electron Microscopy Facilities, Arizona State University). The microscope was loaded with a tungsten film and run at an accelerating voltage of 80keV. High Resolution Electron Microscopy (HREM) images were obtained with a Philips FEI CM-200 Transmission Electron Microscope fitted with a Gatan Orisis CCD camera for image capture. The microscope loaded with a field emission gun (FEG) and used at an accelerating voltage of 200kV. Energy dispersive X-Ray (EDX) spectra were obtained for native ferritin and tungsten-iron apoferritin with a Philips FEI CM-200 mounted X-Ray.

4.6. Selected Area Electron Diffraction

Samples were adsorbed on a holey-copper grid and loaded in a JEOLARM200F aberration corrected scanning transmission electron microscope. The scope was set to diffraction mode and the diffraction patterns were compared to that of a silicon standard at the same camera length to obtain the respective d-spacings.

4.7. X-Ray Diffraction

Samples were first frozen to −80°C overnight, then lyophilized for 24 hours or until dry. Once dry, the powder was dispersed on a zero background holder with 1mm cavity and held together by applying pressure on the powder sample. The holder was loaded on a SIEMENS D5000 powder X-Ray Diffractometer with a fixed horizontal stage. X-Ray radiation was emitted by a CuKα1 source. The range for diffraction detection was run from 2θ values of approximately 10° to 95°. The data was then plotted and indexed with a DIFFRAC Plus measurement software with EVA evaluation software.

4.8. Native Gel Electrophoresis

Samples were buffered in Native Tris-Glycine Sample Buffer (Invitrogen, Carlsbad, CA). 10μl of native ferritin, apoferritin, or WFe loaded apoferritin were loaded into wells of a 4–20% Tris-glycine gel. The gel ran for 6 hours at 125V and 6–12mA/gel. Gels were rinsed for 15min, stained with Simply Blue (Invitrogen, Carlsbad, CA) for 1h, and washed for at least 2h in order to obtain a clear gel background.

4.9. Size Exclusion Chromatography

Samples suspended in water were injected into a Beckman System Gold HPLC mounted with a Phenomenex BioSep-SEC-S4000 column and 0.05M NaCl at pH 7.12 mobile phase. 100μl of sample was injected at a time and absorbance was monitored for at least 35minutes at 280nm and 410nm for elution of protein and metal, respectively.

4.10. SQUID Magnetometry

Samples were freeze-dried using a lyophilizer after freezing at −80°C overnight. Each sample was weighed before measurements. The sample was loaded into a SQUID magnetometer sample holder. A Quantum Design MPMS-5S Superconducting Quantum Interference Device (SQUID) was used for all measurements. Hysteresis curves were obtained at 5K from +30000Oe to −30000Oe. Zero field cooled measurements were obtained by first cooling the sample from room temperature to 2K at zero field. Once the temperature stabilized, a small field of 200Oe was turned on and magnetization measurements were collected upon warm up. Similarly field cooled measurements were performed by cooling the sample to 2K with an on field of 200Oe and measurements were collected during warm up.

Temperature dependent reversal curves were performed at 5K, 10K, 50K, and 150K and the slope of the linear reversals were obtained for further magnetic susceptibility calculations, effective magnetic moment and curie constant calculations. μ_eff was obtained with the formula. μ_eff = 2.82√C_m μ_B.

4.11. Viability and cytotoxicity

A live/dead assay was performed following the instructions in the kit provided by the manufacturer (Biotium, Inc. Hayward, CA). Briefly, 20,000 3T3 cells were seeded onto 24-well plates and cultured for 1 day. The cells were then treated with either 0, 3nM, 17nM, 33nM, or 700nM of WFe apoferritin nanoparticles. The Calcein AM and EthD-III dyes were added to a cell-PBS mixture and allowed to react for 30 minutes. The fluorescence of calcein (530nM emission) and EthD-III (645nm emission) were then measured with a microplate reader. The live/dead percentages were obtained by means of a trypan blue stain. Briefly, 50μl of 0.4%Trypan blue solution was added to 50μl of the cell suspension prior to counting with a hemacytometer. Stained cells were counted as dead and accounted in the live/dead percentage calculations.

4.12. Immunohistochemistry

Following imaging of the live rats, the animal was perfused and the kidneys were extracted and fixed. Following dehydration and section, the kidney was stained with DAPI to visualize cell nuclei. Ferritin was labeled with a rabbit anti-ferritin antibody and with a secondary antibody labeled with Alexa594 (Sigma Aldrich, St. Louis, MO) against rabbit IgG. The immunostained sections were then imaged using a Confocal microscope.