Although nanoparticles synthesized through both physical fabrication and wet chemistry have demonstrated ever increasing diversity and complexity,15
nanoparticles suitable for clinical translation are rare. In this paper, the development of the nanoparticle platform is based on multifunctional fluorescent magnetic nanoparticles incorporating potentially biocompatible components. These nanoparticles each contain an SPIO core, the same material as in the clinical MRI agent Feridex® and the colloidal iron oxide for treating iron deficiency anemia.16
They are coated with a biocompatible siliceous shell, the material component similar to that used in daily Calcium supplements. The fluorescence is from covalently bonded organic fluorophores and FDA-approved fluorophores are available.17
illustrates the general approach in preparing FMN. Starting from SPIO cores, a silanization shell is first grown and then covalently linked to organic fluorophores, along with bio-functional molecules such as drugs or targeting molecules. The fluorescence of the particles permits direct imaging of the particles in a living subject through an intravital microscope, while the magnetic properties are suitable for magnetic targeting. As shown in the transmission electron micrograph of , for nanoparticles deposited onto the TEM grid from a ready to inject solution, FMN contain individual SPIO cores and remain well-dispersed. Dynamic light scattering (DLS) measurement demonstrated that the size of silanized SPIO is 87 nm (supplementary figure 1
). Since FMN made by Cy5.5 fluorophore incorporation, as used in this study, has an emission wavelength interfering with the red laser used for DLS measurement, sizes of FMN-Cy5.5 nanoparticles cannot be directly measured using DLS. However, measurement of FMN samples that incorporate green fluorophores demonstrated that incorporation of fluorophores increased the nanoparticle size by 10 nm, so a rough estimation of the size of FMN-Cy5.5 utilized in this paper is 97 nm. When such nanoparticles are systemically administered in a mouse cancer model, the fluorescence from FMN allows high-resolution fluorescent imaging of the tumor neovasculature. As shown in , no aggregates of nanoparticles in the blood stream were observed using optical microscopy and the FMN fluorescent signal smoothly outlines the tortuous tumor neovasculature, including the very thin capillaries that are only a few microns in diameter.
Schematic illustration of the preparation of the multifunctional nanoparticles with individual SPIO core, a siliceous coating and conjugated fluorophores and biomolecules.
Figure 1 Injectable multifunctional fluorescent magnetic nanoparticles (FMN). A. Transmission electron micrograph showing FMN with individual SPIO core. B. An overlaid image showing the red fluorescence from systemically administered FMN outlining tumor neovasculature, (more ...)
The strong fluorescence from these FMN permits direct observation of their response under an external magnetic control in a living subject. Magnetic targeting can improve specific localization of nanoparticles when sufficient magnetic gradients are applied. This additional applied force can, in principle, help to overcome forces drawing particles away, such as viscous flow forces and other biological barriers1
and help to retain FMN before RES uptake or nonspecific binding, after systemic administration. Magnetic targeting has been developed for many uses,5–8,12
but magnetic targeting of individual nanoparticles with single SPIO cores in living subjects is extremely desirable for using small nanoparticles that can better escape the RES and penetrate tumors. This goal is challenging because a large magnetic field gradient is needed to generate a sufficient magnetic force which requires close proximity between magnets and nanoparticles, since attainable magnetic field gradients fall off very rapidly with distance. To ameliorate this limitation, we implemented an embedded magnetic micromesh that, when magnetized, can generate large magnetic field gradients distributed across a broad spatial range simultaneously. Although the use of two magnetic sources was previously applied for capturing large micron or sub-micron SPIO aggregates,18
such a method has never been exploited to magnetically localize individual SPIO or demonstrated in living subjects. shows the magnitude and distribution of the magnetic field gradient of an electroformed Ni micromesh under a perpendicular magnetic field of 2 KOe, as calculated using Maxwell®
simulation. Ni was chosen as these meshes are readily commercially available to serve as a model for more biocompatible, magnetizable grids. A permanently magnetized mesh might also allow the elimination of the external magnet while retaining the high gradient, but the biocompatibility of permanent magnet materials is less likely than for magnetizable materials. The Ni mesh has a 76 μm pitch, 12 μm wire width and 5 μm wire thickness. The simulation shows that large magnetic field gradients of 104
T/m are generated within 10 μm of the Ni wire edges. Under such large gradients, the magnetic forces exerted on a single 8 nm SPIO cores within this proximity are large enough to overcome the viscous drag on nanoparticles within tumor vasculature (See Supplementary calculation
). The field gradients from the external permanent magnet used to magnetize the mesh are much smaller, 10 T/m, and cannot exert forces large enough to accumulate FMN. The experimental set up is illustrated in . The Ni micromesh is placed directly on the surface of the tumor in the DSC window. The permanent magnet, with field strength of 2 KOe and a maximum field gradient of 10 T/m near the surface, is placed underneath the DSC window and magnetizes both the Ni mesh and FMN in the tumor vasculature.
Figure 2 Maxwell simulation of the magnetic field gradient for the Ni micromesh and magnetic targeting of extravasated FMN. (A) The magnetic field gradient calculated using Maxwell®. The mesh pitch is 76 um and the magnetic field gradient is in T/m. (B) (more ...)
The power of this set up lies in the combined use of two magnetic entities to eliminate the dilemma related to attaining a large magnetic field gradient while maintaining a relatively large distance from a bulky external magnet. With this scheme, the ultrathin magnetizable micromesh can be inserted at the tumor location to generate a large gradient, while the permanent magnet, used to magnetize the micromesh and magnetic nanoparticles, can be placed farther away from the magnetic targeting site. In our experimental set up, the magnetized Ni micromesh generates forces which are large enough to capture FMN. shows that FMN accumulated at the edges of Ni wires near vasculature and that extravasated FMN outline the mesh microstructure with red FMN fluorescence, as expected from the simulated magnetic field gradient distribution of the micromesh. In the control experiment, when only the magnetic mesh was imaged under identical imaging conditions and directly displayed without any intensity adjustment, the signal of the mesh itself is almost negligible. To display the mesh outline from its intrinsic weak fluorescence, as shown in , the image brightness and contrast have to be adjusted, with the inset image showing the weak fluorescence using the intensity scale of . In itself, the fluorescence from the mesh is highly variable over the image, indicating that the fluorescence from brighter regions is not associated with the bare Ni mesh background. On regions of the mesh where few FMN were accumulated, the grid fluorescence is weak, and regions with varying FMN accumulation have variations in apparent width and brightness. All these features indicate that the red color on the mesh in is not from the mesh fluorescence itself, but rather from the accumulated FMN.
This experimental set up is also able to magnetically retain FMN within tumor neovasculature. shows images from three channels that detect FMN fluorescence in red, Angiosense®
(circulating dye) fluorescence in blue, and EGFP tumor signal in green. This figure also shows the outline of regions of interest (ROI) that were specified and tracked for all the imaging frames in three videos, as an aid for quantitative analysis. These videos are provided in the supplementary material
, each contains 60 image frames recorded at 10 s intervals, beginning 4 min after exerting the magnetic force and spanning 10 min. The average intensity of each ROI was calculated using ImageJ and plotted in , where it is clear that the average intensity of FMN signal in the ROI continuously increases throughout the 60 imaging frames and over the 10 min time interval. The real time accumulation of FMN under the magnetic targeting can be directly observed in the supplementary movies
and the increase in average intensity over the ROI comes mostly from an increased fluorescent area, rather than increasing peak intensity. On the other hand, the Angiosense®
and the tumor GFP signals show negligible increases in average intensity over ROI with time. The specific vessel shown here may be at the end of a tumor neovascular sprout, which is often located at the top of the tumor mass and is hence closest to the Ni micromesh and its strong magnetic field gradient.
Figure 3 Magnetic targeting of FMN within tumor neovasculature. (A) Real-time observation of magnetic accumulation of FMN to the mesh edge. The three image channels are: red for FMN fluorescence, blue for Angiosense-750 fluorescence, and green for EGFP-transfected (more ...)
Successful magnetic retention of FMN can also be observed within the well-developed neovasculature. In , the image channels for Angiosense® (left panel, blue color) and FMN (right panel, red color) are placed side by side for comparison. For tumor neovasculature at the center of the image, which is outlined in a relatively consistent manner by Angiosense®, punctate nodal spots at edges of Ni wires are obvious in the FMN signal channel. This is because the strong magnetic forces from the edges of Ni wires are able to retain FMN, resulting in stronger fluorescence signal at the nodal spots. The ratio of averaged FMN fluorescence intensities for a ROI that encircles the nodal spot at the Ni wire edge versus that from a ROI that encloses the vessel area near the center of the Ni mesh hole is much larger than that for the Angiosense® channel. A few pairs of such locations are indicated using colored arrows in . Corresponding intensity ratios for each pair are shown in the table of , along with a bar graph of the average values and standard deviations. The average intensity ratio for FMN signal is five times that of Angiosense® signal. This clearly demonstrates that FMN were able to respond to the external magnetic field gradient generated by the Ni micromesh and thus magnetically accumulated close to the wire edge to form nodal spots with intense fluorescence.
The magnetic targeting of FMN to the tumor region could largely impact the functions of biomolecules conjugated to FMN surfaces. Here we used RGD binding to tumor angiogenic target αv
integrin as a model system, which is reported to cause apoptosis of tumor blood vessels and promote tumor regression.13
The scrambled form RAD (Arg - D - Ala – Asp)19
was utilized in control experiments. FMN-RGD demonstrated binding specificity to cultured U87MG human glioblastoma cells, as compared to FMN-RAD (). For experiments in living subjects, we selected the intensity of tumor EGFP signal to monitor the RGD-induced effects in different experiments since nanoparticles could extravasate out of tumor leaky vessels and remain in tumor regions for extended time regardless of surface molecular specificity, in line with the enhanced permeability and retention effect of nanoparticles.20, 21
shows that intravenously injected FMN-RGD caused tumor regression (n=3), while tumors continued to grow after FMN-RAD injection (n=2). For FMN-RGD injection together with magnetic targeting, tumors regressed at a much faster pace (tumor signal decay half-life of 0.853 days) than when injecting FMN-RGD without any magnetic force with a tumor signal decay half-life of 6.197 days (n=3, p<0.05, and Supplementary Figures 2, 3
). The tumor signal decay pattern under magnetic targeting is similar to that obtained by doubling the FMN-RGD dosage (Supplementary Figures 2 and 3
), suggesting that expedited tumor regression under magnetic targeting is due to the retention of more FMN-RGD in the tumor region by virtue of magnetic targeting. More rigorous dosage experiments will need to be eventually performed to exactly quantify the dose equivalence of the magnetic targeting scheme.
Figure 4 Cell staining experiments demonstrate binding specificity of FMN-RGD to cultured U87MG human glioblastoma cells in comparison with FMN-RAD. U87MG cells over-express αvβ3 integrin on the surface that bind to RGD. U87MG cells are transfected (more ...)
Figure 5 RGD-conjugated FMN in combination with external magnetic control expedites tumor regression in a U87MG human glioblastoma xenograft mouse model. (A–C) EGFP-transfected tumor image channels show tumor intensity change within days of imaging for (more ...)