Macrophages are considered an important component of the immunological defense system (1). They provide both innate and acquired immunity and protect the host from invading pathogens. Conversely, due to their function in the development of inflammatory response, macrophages are associated with inflammatory diseases such as tuberculosis (2), legionellosis (3), rheumatoid arthritis (4, 5), cirrhosis (6), and atherosclerosis (1, 7). Imaging macrophages in these lesions may allow us to assess the status of inflammation and disease.
Among diseases in which macrophages are involved, atherosclerosis caused by plaques is one of the most important targets for medical diagnosis and treatment because coronary artery disease is one of the major causes of death in western countries. Accumulating evidence indicates that macrophages take part in the initiation of plaques to the state that ultimately leads to rupture (7, 8). Consequently, macrophages are an appropriate target to evaluate the inflammatory status of plaques (9, 10). Recent advances in magnetic resonance imaging (MRI) technology has allowed for the noninvasive visualization of atherosclerotic plaques. For example, ultrasmall superparamagnetic iron oxide (USPIO) contrast agents, such as ferumoxtran-10 (Combidex®), have been evaluated as potential MRI contrast agents to detect macrophages within plaques (8–13). Not only in vitro but also in vivo studies warrant potentiality of USPIOs to evaluate the plaque condition, because it has been confirmed that USPIOs which administrated systemically in vivo accumulate in macrophage in atherosclerotic plaque of rabbits and human (8, 11). However, their uptake efficiency into macrophage may be suboptimal (9, 10).
We have recently shown that protein cage architectures, such as virus capsids (14, 15), ferritins (15, 16), and heat shock proteins (17), are useful templates for the synthesis of ferrimagnetic nanomaterials encapsulated within the interior cavity of the protein cage. These nano materials are comparable in size to USPIO contrast agents but possess unique features not present in the commercially available USPIO contrast agents. This includes the fact that protein cage architectures are precise assemblies of subunits with an extremely homogeneous size distribution (18). The homogeneity of the protein cage template leads to homogeneity of the templated iron oxide nanoparticles within the protein cage architecture. Furthermore, protein cage templates possess three distinct interfaces, that is, interior surface of the cage, exterior surface of the cage, and the interface between subunits, which can be chemically and/or genetically modified to impart novel function by design to the template including targeting and therapeutic delivery (18). In addition, these highly symmetrical protein cage architectures provide an ideal multivalent platform for attachment and presentation of Gd(III) binding chelates to generate materials with very high R1 relaxivity (19–23). These features suggest that protein cage nanoparticle composites have the potential to serve as a new class of T1 and T2 MRI contrast agents.
The aim of this study was to characterize the MRI properties of a ferritin protein cage-iron oxide nano-composite material and investigate its use as an USPIO MRI contrast agent to label macrophages.
MATERIALS AND METHODS
Cloning and Purification of Recombinant Human H Chain Ferritin (rHFn)
The cloning, heterologous expression, and purification of recombinant human H chain ferritin (rHFn) has been described previously (16). Briefly, the gene encoding the HFn was amplified by polymerase chain reaction (PCR) followed by cloning into the NdeI/BamHI restriction endonuclease sites of the pET-30a(+) (Novagen, Madison, WI) for expression of the full-length HFn protein. The HFn clone was expressed in and purified from E. coli (BL21 (DE3): Novagen). Cells were harvested by centrifugation and subsequently resuspended in 45 mL of lysis buffer (100 mM HEPES, pH 8.0, 50 mM NaCl, 50 μg/mL of Lysozyme, 60 μg/mL of DNAse, 100 μg/mL of RNAse). The suspension was incubated for 30 min at room temperature followed by cell disruption by a combination of French Press lysis and sonication on ice. The solution was centrifuged to remove cell debris and the resulting supernatant heated at 60°C for 10 min to precipitate many of the E. coli proteins, which were subsequently removed by centrifugation. The resulting supernatant was subjected to size exclusion chromatography (SEC: Amersham-Pharmacia, Piscataway, NJ) using a Superose 6 column to isolate intact assembled rHFn protein cages. The purified rHFn protein cages were further characterized by UV/Vis spectroscopy, dynamic light scattering (DLS: Brookhaven, 90Plus particle size analyzer) and transmission electron microscopy (TEM: LEO 912AB).
Iron Oxide Mineralization and Characterization
Iron oxide mineralization constrained within the rHFn protein cage structure was performed under three different iron loading factor conditions, that is, 1000Fe, 3000Fe, and 5000Fe per cage, respectively. Briefly, 10 mL solution of the rHFn (10 mg, 19.8 nmol in 100 mM NaCl) was added to a jacketed reaction vessel under a N2 atmosphere. The temperature of the vessel was brought up to and maintained at 65°C by circulating water through the jacketed flask. The pH was titrated to 8.5 with 50 mM NaOH (718 Auto Titrator, Brinkmann). Ammonium iron(II) sulfate hexahydrate ((NH4)2Fe(SO4) · 6H2O) and hydrogen peroxide (H2O2) were used as an iron source and oxidant, respectively. To achieve a theoretical Fe loading factor of 1000Fe, 3000Fe, and 5000Fe per protein cage, 1580, 4741, and 7901 μL of (NH4)2Fe(SO4) · 6H2O solution (12.5 mM) were used in independent reactions, respectively. Because the mineralization reaction proceeds according to reaction 1, stoichiometric equivalents of freshly prepared degassed H2O2 (4.17 mM) was also added as an oxidant. The (NH4)2Fe(SO4) · 6H2O and H2O2 solutions were injected simultaneously into the reaction vessel at a constant rate of 158 μL/min (100Fe/(cage · min)) using a syringe pump (Kd Scientific). After the mineralization procedure, 200 μL of 300 mM sodium citrate was added to chelate any remaining free Fe, and the sample was dialyzed against Dulbecco's Phosphate Buffered Saline (DPBS). The mineralized rHFn was purified from any aggregation products and analyzed by SEC with Superose 6 while monitoring absorbance at 280 nm and 410 nm for elution of protein and mineral, respectively. Only mono-dispersed mineralized rHFn was used for subsequent experiments.
MR Relaxivity Measurements and MR Imaging of the Mineralized rHFn
Each sample was serially diluted from 8 mM to 0.1 mM (Fe concentration) in DPBS. One hundred microliters of each solution was added in a 96-well PCR plate (E&K Scientific Products, Inc., Santa Clara, CA) and embedded in 1% agarose supplemented with 1% CuSO4 (24). Measurements were performed using a 1.5 Tesla (T) whole-body scanner (GE Healthcare, Waukesha, WI) equipped with a 5-inch receive-only surface coil. To measure longitudinal relaxation times (T1), an inversion recovery spin echo (IR-SE) sequence was used (TR = 6000 ms TE = 9 ms, TI = 50, 100, 150, 200, 250, 300, 350, 400, 450, 550, 650, 750, 850, 950, 2000, and 3000 ms, matrix = 128 × 128, slice thickness = 3.0 mm, NEX = 1, field of view [FOV] = 12 cm). For the measurement of transverse relaxation times (T2), a spin-echo sequence with different echo times was used (TR = 3000 ms TE = 9, 12, 20, 30, 50, 80, 100, and 160 ms, matrix = 192 × 160, slice thickness = 3.0 mm, NEX = 1, FOV = 12 cm). The T1 for each sample was calculated in MATLAB (The Mathworks, Natick, MA) by using the fminsearch routine to perform a nonlinear fit of the data to the inversion-recovery signal equation. The T2 for each sample was also calculated using fminsearch, but by fitting to a mono-exponential decay curve. Relaxivity, R1 and R2, of each sample were calculated using the T1 and T2 data and Fe concentration determined by an inductively coupled plasma mass spectrometry (ICP-MS: Agilent, Santa Clara, CA). To assess the relative visual signal at 1.5T, the images were acquired by using a gradient echo (GRE) sequence (TR = 100 ms TE = 10 ms, Flip angle = 30°, matrix = 256 × 256, slice thickness = 3.0 mm, NEX = 1, FOV = 12 cm).
In Vitro Contrast Agent Uptake Assay
Murine macrophage cells (RAW) were cultured using Dulbecco's Modified Eagle Medium (DMEM, cat. 11995-065, Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum, 100 units/mL of penicillin, 100 μg/mL of streptomycin at 37°C in 5% CO2 atmosphere. Cells were passaged every 2–3 days.
Uptake of the mineralized rHFn with various Fe loading factors was evaluated using ICP-MS and Prussian blue staining under two different conditions, that is, constant rHFn concentration and constant Fe concentration. In the case of the constant rHFn condition, the mineralized rHFn was added to a final concentration of 0.3 mg/mL (equivalent to 33 μg Fe/mL, 100 μg Fe/mL and 165 μg Fe/mL for 1000Fe, 3000Fe, and 5000Fe per rHFn cage, respectively). In the case of constant Fe concentration, the mineralized rHFn was added to a final concentration of 165 μg Fe/mL (equivalent to 1.5 mg rHFn/mL, 0.5 mg rHFn/mL, and 0.3 mg rHFn/ mL for 1000Fe, 3000Fe, and 5000Fe per HFn cage, respectively). For comparison, cells were also incubated with 165 μg Fe/mL of Fe based MR contrast agent, ferumoxides or ferumoxtran-10 (i.e., Feridex® and Combidex®, respectively, AMAG Pharmaceuticals, Inc., Cambridge, MA).
For the analysis using ICP-MS, cells were seeded into six-well polystyrene plate at the concentration of 1 × 106 cells/ well. After 24 h of incubation, the medium was removed and 1.6 mL of a solution containing a mixture of the fresh cell culture medium and the mineralized rHFn samples at a ratio of 3:1 was added to a well. The cells were further incubated with the mineralized rHFn at 37°C in 5% CO2 atmosphere for 24 or 72 h. After incubation, the solution was aspirated and the cells gently washed two times with DPBS followed by adherent cell removal and collection using 0.25% trypsin-EDTA (25200-056, Invitrogen). The number of viable cells was determined by trypan blue exclusion assay and counting using a hemocytometer. To quantify the amount of Fe taken up by the cells, cell pellets were dissolved in nitric acid and subjected to ICP-MS analysis. The data were assessed using a Student's t-test.
For analysis of Fe uptake by Prussian blue staining, cells were cultured in a four-well slide glass chamber in the same manner as described above. Subsequently, the cells were fixed with 4% of formaldehyde then washed with distilled water three times. Five percent of potassium ferrocyanide was mixed with 5% hydrochloric acid at a ratio of 1:1 just before use for staining. Fixed cells were incubated with the mixed solution for 30 min at room temperature, washed by distilled water three times, and counterstained with Nuclear Fast Red.
MRI of Macrophage Cells
Macrophages were incubated with 165 μg Fe/mL of the mineralized rHFn, ferumoxides or Ferumoxtran-10 for 24 h in the same manner as described above for the uptake assay. Then, 1 × 106 cells were suspended in 100 μL of DPBS, transferred in a 96-well PCR plate and allowed to settle by gravity. The MR imaging of the cells was performed at 1.5T with the same setup described above using a GRE sequence (TR = 100 ms TE = 10 ms, Flip angle = 30°, matrix = 256 × 256, slice thickness = 1.0 mm, NEX = 1, FOV = 12 cm).
Characterization of Mineralized rHFn
Recombinant human H chain ferritin (rHFn) was used as an active, constrained reaction template for the synthesis of ferrimagnetic iron oxide nanoparticles for application as an MRI contrast agent. Analysis by TEM demonstrated that rHFn has a cage-like architecture of approximately 12-nm diameter, which is identical to native mammalian ferritins (Fig. 1) (25). Dynamic light scattering analysis also demonstrated that the rHFn protein cage is monodispersed (mean size = 13.5 nm), suggesting that the heterologously expressed rHFn is assembled from 24 identical subunits to form a spherical architecture with 4:3:2 symmetry.
The spatially selective synthesis of an iron oxide nanoparticle within the rHFn protein cage architecture is supported by multiple lines of evidence. In the presence of the purified rHFn, a dark brown solution formed with no visible precipitation, whereas in protein-free control reactions there was an immediate formation of a bulk precipitate. TEM analysis of nonstained rHFn samples, after the iron oxide synthesis reactions, revealed electron dense nanoparticles with narrow size distributions (Fig. 2). The average size of the particles increased from 3.6 to 5.9 nm with increasing Fe loading factor from 1000Fe to 5000Fe per cage. Indexing of selected area electron diffraction patterns revealed that these particles were crystalline material. The measured d-spacing of the particles was compared with the Joint Committee on Powder Diffraction Standards (JCPDS) and confirmed to be ascribed to either maghemite (γ-Fe2O3) or magnetite (Fe3O4) (Fig. 3). Due to the similarity in lattice constants for these two cubic structures (a= 8.352 of maghemite or a= 8.396 of magnetite, both from JCPDS), it is difficult to unambiguously distinguish them from the electron diffraction patterns alone. Further structural characterization of the mineralized rHFn is not significant in this study because the magnetic properties of maghemite and magnetite are similar due to their similarity of crystal structure. The theoretical core diameters calculated for a uniform spherical particle of the cubic iron oxide (either Fe3O4 or γ-Fe2O3) phase at loading factor of 1000Fe, 3000Fe, and 5000Fe are 3.6 nm, 5.2 nm, and 6.2 nm, respectively. The observed mean diameter of the iron oxide core formed inside of the HFn are close to the theoretical value (Fig. 2). This suggests that almost all Fe added during the reaction are accumulated inside of the HFn cage. This is favorable character of HFn for template iron oxide core formation. Because interior cavity of HFn in which iron oxide is formed is approximately 8 nm in diameter (25), it is large enough to accumulate Fe as an iron oxide even at the loading factor of 5000Fe/cage. SEC analysis of the mineralized rHFn demonstrated the co-elution of the protein cage and the iron oxide indicating the composite nature of rHFn - iron oxide nanoparticle (Fig. 4a). Furthermore, both the pre- and postmineralized rHFn had identical SEC retention times, indicating that the overall size of the protein cage had not been altered by the iron oxide synthesis reaction (Fig. 4a).
However, a smaller peak on the leading edge of the SEC profile after the mineralization is probably due to aggregated rHFn during the mineralization process but was easily separated by chromatography. DLS analysis of the mineralized rHFn after SEC purification indicates that the protein cage is approximately 14 nm in diameter, which is nearly identical to intact rHFn before mineralization (Figs. 1b, 4b). In combination, these results demonstrate that the synthesis of magnetite (or maghemite) nanoparticles occurred specifically within the interior cavity of the rHFn protein cage without significantly perturbing the overall protein cage architecture.
MR Relaxivity Measurement and Imaging of the Mineralized rHFn
To evaluate the potential of the iron oxide mineralized rHFn as an MRI contrast agent, MR relaxivity determination and imaging of the synthesized materials was performed (Fig. 5). As shown in Figure 5, the T2* signal loss became larger with increasing Fe loading factor per cage under the same Fe concentration. The image of HFn5000Fe was similar to that of ferumoxides and ferumoxtran-10. Likewise, R1 and R2 relaxivities of the mineralized rHFn increased with increasing Fe loading factor. The R1 and R2 of HFn5000Fe were comparable to that of ferumoxides and ferumoxtran-10. These results indicate that mineralized rHFn protein cages could have similar T1/T2* MR imaging properties to known iron-oxide contrast agents. Previous measurements on magneto-ferritin having iron oxide cores of 7.3 ± 1.4 nm diameter reported higher values of R2 (26) due to the larger particle size as compared with this study (3.6 ± 0.7 nm, 5.1 ± 0.9 nm and 5.9 ± 0.9 nm). This strongly supports the concept that these ferritin cages act as reaction vessels with significant control over particle size and associated magnetic properties.
Uptake Assay and MR Imaging of Macrophage Cells
The mineralized rHFn cages were taken up by the macrophages regardless of Fe loading factor per cage. As the Prussian blue staining results show, mineralized rHFn cages were phagocytosed by the RAW cells to a similar degree as ferumoxides (Fig. 6) when the same Fe amount was added. On the other hand, only limited staining was observed in the cells incubated with ferumoxtran-10, even though the cells were treated with the same amount of Fe as the cells treated with other contrast agents.
Quantitative analysis of Fe uptake determined by ICP-MS (Fig. 7) showed results similar to the Prussian blue staining. Cells incubated at constant rHFn concentration, with 1000Fe, 3000Fe, and 5000Fe per cage accumulated 18 pg/cell, 52 pg/cell, and 116 pg/cell of Fe after 72 h of incubation, respectively. This suggests that the RAW cells phagocytosed nearly identical amounts of rHFn protein cages, irrespective of Fe loading factor per cage. Even when the RAW cells were incubated at fixed Fe concentration, the cells took up more Fe with increasing Fe loading factor in the cage they treated with (21 pg/cell, 71 pg/cell, and 116 pg/cell after 72 h of incubation, respectively). These results suggest that loading more Fe per cage is advantageous not only from large T2* signal loss but also effective Fe uptake by macrophage. At equivalent Fe concentration, the amount of Fe taken up by the cells cultured with the mineralized protein cage was less than that of cells cultured with ferumoxides. Ferumoxide incubation resulted in 152 pg Fe/cell after 72 h, and statistical comparison with Fe uptake from ferritin incubation yielded P = 0.03, 0.08, and 0.3 for HFn1000Fe, 3000Fe, and 5000Fe, respectively. However, significantly more Fe was accumulated in cells incubated with rHFn than cells cultured with ferumoxtran-10. Ferumoxtran-10 incubation resulted in 3 pg Fe/cell after 72 h, and statistical comparison with Fe uptake from ferritin incubation yielded P = 0.02, 0.005, and 0.009 for HFn1000Fe, 3000Fe, and 5000Fe, respectively. Cell viability data for cells only vs. cells incubated with the various contrast agents (Fig. 8) did show some decrease in cell count for the contrast agents but this was not statistically significant (P > 0.05).
Labeled macrophages showed a wide range of T2* signal loss by MRI whereas cells without any contrast had no signal loss under the same total Fe concentration (Fig. 9). The signal loss from incubated cells became larger across the agents ferumoxtran-10, rHFn 1000Fe/cage, 3000Fe/cage, 5000Fe/cage, and ferumoxides. It should be noted that although the R2 relaxivities of rHFn 1000Fe/cage, and rHFn 3000Fe/cage were less than that of ferumoxtran-10, the MRI signal loss of cells labeled with the mineralized rHFn was greater than that of the cells labeled with ferumoxtran-10. This is likely due to greater uptake of rHFn protein cages by macrophages in vitro than ferumoxtran-10.
We have demonstrated the potential of using a biomimetic approach for developing a new class of MRI contrast agents using a protein cage, human ferritin, to image macrophages in vitro as the first step to the assessment of inflammatory diseases, such as atherosclerosis. The natural role of ferritin is the sequestration of Fe as a hydrated ferric oxide (25, 27), and it is, therefore, an appropriate reaction vessel for ferrimagnetic iron oxide synthesis. In the present study, rHFn was exploited as a template for superparamagnetic iron oxide nanoparticle synthesis for use as an MRI contrast agent. The protein cage mediated iron oxide nanoparticles were compared with iron oxide MR contrast agents, such as dextran-coated ferumoxides and ferumoxtran-10, to evaluate cell uptake and the MR signal properties.
The important findings of this study are that the mineralized protein cages are readily taken up by macrophages in vitro and provide strong T2* MR contrast. This suggests that this material has a great potential as an MRI contrast agent to assess inflammatory status such as atherosclerotic plaque progression/regression. At equivalent iron concentrations, ferumoxides are taken up by the macrophages more than the mineralized rHFn using an in vitro cellular uptake assay (Figs. 6, 7). However, likely due to their larger size, ferumoxides are trapped by Kupffer cells in liver tissue during first-pass after injection (28, 29), reducing their circulation time, making them less useful for macrophage and atherosclerosis imaging in vivo. On the other hand, ferumoxtran-10 can reach atherosclerotic lesions in vivo (8, 11), likely due to a longer circulation time. These materials have been examined as potential contrast agents to image macrophage-rich plaques. However, their uptake efficiency into macrophage may be suboptimal. The overall size of the mineralized rHFn protein cages (14 nm, Fig. 1b) is closer to that of ferumoxtran-10 (29.5 ± 23.1 nm) (30) rather than that of ferumoxides (58.5 ± 185.8 nm) (30). Nevertheless, cells incubated with the rHFn protein cages took up statistically significantly more (P < 0.05) iron than those incubated with ferumoxtran-10 (7- to 39-fold more) at the same Fe incubation concentration, and cell uptake levels were fairly comparable to the ferumoxides. This indicates that the protein cage based contrast agents have significant potential for labeling macrophages within inflammatory lesion such as atherosclerotic plaques, warranting in vivo evaluation. The incorporation of iron oxide nanoparticle within rHFn can be controlled precisely with a very narrow size distribution (Fig. 2). Because the size of a magnetic particle is a significant factor, with a large effect on the MRI properties, precise control of the iron oxide particle size may be advantageous.
Another significant advantage of the cage-based material may be that the protein cage can serve not only as a template for iron oxide nanoparticle synthesis but also as a platform for imparting additional functionality such as cell-specific targeting capability (16, 31, 32). Protein cages such as ferritin are self-assembled from a limited number of protein subunits into precisely defined architectures. The exterior and interior surfaces of a cage possess chemically addressable functional groups. Genetic engineering techniques are also available for varying amino acid patterns of a subunit or introducing functional peptide sequences. We have previously demonstrated the introduction of cancer cell targeting moieties or therapeutic agents to several different protein cages (16, 31, 33, 34). It has been reported that “activated” macrophages in an unstable plaque express specific receptors such as a chemokine-receptor, CCR2 (35). If a specific targeting moiety for “activated” macrophages is introduced to the mineralized rHFn, the resultant material might be expected to localize in macrophages within unstable atherosclerotic plaques and serve as a superior diagnostic test to evaluate the inflammatory state or vulnerability of the plaque to rupture. Furthermore, if a therapeutic agent can be co-loaded into the protein cage, the composite material could simultaneously achieve therapeutic effects on atherosclerotic plaques as well as imaging capability.
As described above, the mineralized rHFn is believed to have great potential as a MRI contrast agent. However, it is important to investigate biocompatibility of the material before proceeding to clinical studies. Although the results of the cell viability assay do not show any statistically significant toxicity of the mineralized rHFn compared with Ferumoxides and Ferumoxtran-10, it does show some decrease in cell count. More extensive in vitro and in vivo toxicity investigations are needed in the future and, even though we have used human recombinant protein, it is also critical to address the immunogenicity of the mineralized ferritin. These investigations are currently under way.
In conclusion, iron in the form of a synthetically mineralized magnetite (or maghemite) nanoparticle, encapsulated within the protein cage architecture of a recombinant human H chain ferritin (rHFn) shows great promise as an MRI contrast agent. The biotemplating by the ferritin cage results in mineral cores with narrow size distribution and very homogeneous properties. The macrophage uptake in vitro and T2* MRI properties of the mineralized rHFn compare favorably with known iron oxide MRI contrast agents. This composite material is expected to have great potential as a MRI contrast agent to assess the state of macrophage-rich atherosclerotic plaques in vivo. Future work will be directed toward incorporating cell specific targeting moieties to the rHFn for cell-specific in vivo applications.
The authors thank AMAG Pharmaceuticals, Inc., for providing Ferumoxtran-10 (Combidex®).