Magnetic Nanoparticles as Mediators of Ligand-free Activation of Egfr Signaling

Background: Magnetic nanoparticles (NPs) are of particular interest in biomedical research, and have been exploited for molecular separation, gene/drug delivery, magnetic resonance imaging, and hyperthermic cancer therapy. In the case of cultured cells, magnetic manipulation of NPs provides the means for studying processes induced by mechanotransduction or by local clustering of targeted macromolecules, e.g. cell surface receptors. The latter are normally activated by binding of their natural ligands mediating key signaling pathways such as those associated with the epidermal growth factor (EGFR). However, it has been reported that EGFR may be dimerized and activated even in the absence of ligands. The present study assessed whether receptor clustering induced by physical means alone suffices for activating EGFR in quiescent cells.


Introduction
Nanoparticles differing in composition, shape, size, and intrinsic optical, electronic and magnetic properties have been used in diverse biological applications such as imaging, sensing and separation [1,2,3,4]. In particular, magnetic NPs [5] have been exploited for molecular separation, gene/drug delivery, and magnetic resonance imaging [6,7]. As sensors and actuators they have been used to sense femtomolar concentrations of proteins, mRNA or viruses [8], for focused heat-induced manipulation of ion channels [9], or for mechanotransduction of ion channels in neurons [10]. Some cell surface receptors are activated by clustering, a prominent example being the FceR1 receptor on basophils and mast cells that is aggregated upon recognition of multivalent allergens by bound IgE [11]. Mannix et al. demonstrated that monovalent antigen attached to SPIONs could induce mast cell activation, manifested by Ca 2+ waves arising after clustering the FceR1 by a magnetic field [12]. Apoptosis of tumor cells has been achieved by magnetic aggregation of SPIONs coupled to a monoclonal antibody against DR4 receptors [13], although it was necessary to apply the magnetic field for 2 hr in order to observe caspase 3 activity. The same group achieved a similar result in live zebrafish embryos by targeting the ovarian TNF receptor with microinjected SPIONs and applying a field for 24 or 48 h. A number of recent studies have utilized large magnetic NPs introduced by microinjection to redistribute materials inside cells. Examples are cytoskeletal reorganization induced by Raf1 NPs [14] and microtubule assembly in Xenopus oocyte extracts by RANQ-GTP coupled NPs [15].
The epidermal growth factor receptor (EGFR, ErbB1, HER1), a prototypic transmembrane tyrosine kinase receptor, is a member of the ErbB (HER) family. Ligand binding results in dimerization and subsequent trans-phosphorylation of several tyrosine residues in the intracellular C-terminal tail of the receptor [16,17,18]. The adaptor proteins Shc, Grb2 and Cbl recognize these phosphotyrosines, thereby propagating downstream signaling, effector functions and receptor internalization [19,20]. These signaling cascades orchestrate a wide range of cellular processes such as cell differentiation, motility, and cell division [21,22].
It has not been firmly established whether receptor dimers or oligomers can be activated and initiate downstream signaling in the absence of physiological ligands. Yu et al. [23] reported that EGFR dimerized and was activated merely by association with a2b1 integrins in serum deprived cells while Takahashi et al. [24] studied the effect of extracellular matrix glycans on ligand free activation of ErbB3 mutants. However, another investigation of integrin association by Alexi et al. [25] failed to demonstrate EGFR activation without added ligand, and the authors concluded that autocrine activation of the receptor was likely to have occurred in some of the other studies. Monoclonal antibodies that block ligand binding inhibit EGFR signaling and some cause down regulation of the receptor [26,27,28], suggesting that ligand binding is indeed required for EGFR activation. Some of these antibodies have been humanized and used to treat cancers expressing high levels of EGFR [29,30,31]. Certain mutations of EGFR cause a constitutive ligand independent activation of the receptor and such forms often arise under conditions favoring cellular transformation [32,33].
We initiated the present study to determine whether clustering of EGFR on living cells could bypass the ligand requirement for activating the receptor. We exploited the biocompatibility, tunable surface properties and ease of preparation of superparamagnetic iron oxide NPs (SPIONs), employing the latter as magnetic actuators (''switches''). Non-ligand mediated clustering activated the EGFR and led to internalization of both the receptor and the SPIONs.

Results
The properties of magnetic nanostructures depend on size. Below a certain critical dimension they exhibit the phenomenon of superparamagnetism, which is characterized by a high magnetic susceptibility and the absence of a residual field [3,34,35]. SPIONs were synthesized by alkaline co-precipitation using citrate as a stabilizing agent [34,35]. Transmission electron microscopy (TEM) images indicated the presence of well-defined, spherical shaped particles 12-15 nm in diameter (Figure S1a-e). A more homogenous SPION population was obtained by magnetic fractionation followed by functionalization with streptavidin (strv-SPION) (see Materials and Methods).

Magnetic Switches are Created by Covalently Attaching
Cell-targeting Molecules to SPION Figure 1 shows pictorially the composition of ''magnetic switches'' (MS) and the experimental approach for specific targeting to EGFR. MS were created by coupling strv-SPION to a biotinylated anti-EGFR (ectodomain) monoclonal antibody (528 MAb). This antibody blocks binding of the normal ligands for EGFR, and thus inhibits ligand-induced activation as well as cell proliferation and orthotopic tumor growth [36,37]. We adjusted the reagent concentrations to achieve #1 streptavidin and ,1 MAb per nanoparticle. We also demonstrated that strv-SPIONs by themselves can act as MS if targeted to cells expressing a chimeric EGFR with a biotinylated acyl carrier protein (ACP) tag on the amino terminus of the receptor (see Materials and Methods). In order to visualize the MS on the cell surface by fluorescence, unoccupied biotin binding sites on strv-SPION were loaded with biocytin-Alexa 488 or biocytin-Alexa 546.

SPIONS Aggregate and Cause Activation of the Membrane-bound EGFR Only in the Presence of a Magnetic Field
Application of the targeted MS to A431 cells and washing to remove unbound particles resulted in MS distributed over the entire cell surface ( Figure S2a-d). Untargeted SPIONs (lacking anti-EGFR) did not bind to the cells ( Figure S2e). Exposure to an external magnetic field led to aggregation of MS into small clusters ( Figure 2a) and activation of EGFR detected by an EGFR-specific anti-phosphotyrosine (pY) antibody (Fig. 2a9). In contrast, no EGFR-specific pY signal was observed in cells that were not exposed to a magnetic field after binding of MS (Figure 2b,b9). Figure 2c shows a 2-dimensional colocalization histogram of EGFR phosphorylation and the MS signals from the images in Figure 2a,a9, revealing a strong, positive correlation (colocalization parameters: Manders coefficient 0.78, overlapping coefficient 0.82, and correlation coefficient 0.80).
In order to define the dipolar interactions and the anticipated extent of magnetic field induced clustering of MS bound to EGFR on the cell surface, we carried out a detailed theoretical analysis of the experimental system. Finite element analysis calculations indicated that the magnetic field experienced by MS on the cell surface would be close to saturating strength ( Figure S3). At the separation distance of 10 nm the calculated force between neighboring MS on cell surface was , 0.25 pN, decaying exponentially with increasing separation distance. We conclude that in our experimental system the strengths of the magnetic dipole moment and of the external field sufficed for inducing the MS to form stable clusters on the cell surface (Supporting Information S1 and Figure S4).
Scanning electron microscopy (SEM) resolved the local clustering of MS targeted to EGFR on A431 cell membranes. In the absence of a magnetic field, MS were distributed densely on the cell surface but were individually distinct with little or no clustering (Figure 2d,d9). In contrast, cells treated with MS in the presence of a magnetic field showed a dense population of particles arranged in local clusters of , 100-200 nm (Figure 2e,e9).
MS did not exhibit magnetic hysteresis at room temperature, indicating the existence of superparamagnetic behavior ( Figure  S3). Based on the magnetic properties, we calculated the magnetic diameter of MS to be , 10 nm (see Supporting Information S1), a result which was corroborated by TEM and the estimation of the hydrodynamic diameter as ,15 nm by dynamic light scattering (Fig. S1).

EGFR Activation is Dependent on the Time of Magnetic Exposure
To determine the effect of magnetization as a function of time, MS targeted to EGFR in A431 cells were exposed to a magnetic field for various time periods and the resulting pY-EGFR signals were measured by image analysis (Figure 3a,b,c) and western blotting (Figure 3d,e). Exposures as short as a few seconds showed receptor activation whereas no pY-EGFR was detected in the absence of a magnetic field. Distinct pY-EGFR signals were observed and increased for up to 3 min of magnetization (Figure 3b,c) but then remained constant, indicating that maximum activation had been achieved with the given field. Quantification by western blot analysis ( Figure 3 d.e) indicated similar signal intensities for pY-EGFR at saturating concentrations of natural ligand and after a 3 min exposure of the MS to a magnetic field. These observations were consistent with measured magnetization data for the SPIONs ( Figures S3 and S4 and Supporting Information S1).

Magnetic Switches Induce Downstream Signaling of the EGFR
To determine whether magnetism-induced activation of the EGFR triggers the same signal cascade as that induced by ligand binding, we investigated the recruitment, redistribution and state of phosphorylation of proteins in the MAP Kinase pathway. Shc (Src homology and collagen) is an early adapter protein that is recruited to the plasma membrane after EGF stimulation, where it binds specific phosphotyrosines in the Cterminus of the activated receptor and is itself phosphorylated by the receptor [38,39]. We induced activation of EGFR by magnetization of tagged MS and assayed by immunofluorescence the appearance as well as the colocalization of p317-Shc with the SPIONs after further incubation of the cells for 10 min at 37 uC, compared with cells not exposed to the magnetic field ( Figure 4 a,b,c). The two-dimensional colocalization histograms are shown for each condition and were derived after deconvolution of image stacks using SVI Huygens software. The Manders coefficient was 0.95 after 30 s and 0.98 for 180 s of magnetization, respectively. Subsequent to signaling by EGF at the plasma membrane, the receptor was endocytosed through clathrin-coated pits [40,41]. We observed this phenomenon after activation by SPIONs, which colocalized with the early endosomal marker EEA1 in cells transduced magnetically for 3 min and then incubated for 15 min at 37uC (Figure 5a and b). In the absence of the magnetic field the MS remained dispersed on the cell surface and there was no activation (phosphorylation) during incubation at 37uC, whereas MS clustering and activation of EGFR was evident after application of the magnetic field for 30 s and subsequent incubation at 37uC ( Figure S5 a-d). Simulations for the cluster formation dynamics and Brownian relaxation for such particles are given in the Supporting Information S1 and Figure S6. High Order Clustering of the EGFR is Required for Activation by MS Bivalent crosslinking of biotinylated 528 MAb bound to EGFR mediated by streptavidin-Atto565 ( Figure S7) did not result in a detectable pY-EGFR signal, indicating that protein bridging between the MAb bound to the receptor was incapable of associating the EGFR with the kinase domainsoriented in an active conformation. We conclude that higher order clustering of the EGFR and/or a more intimate contact of the receptors are required and can be achieved only by application of the magnetic field.

Magnetic Activation of EGFR by Targeted SPIONs does not Depend on High Expression Levels of the Receptor on the Cell Membrane or on Targeting by a Large Monoclonal Antibody
To determine whether MS were effective only on cells with very high densities of receptor, we performed the same experiments on Hela cells expressing 5?10 4 EGFR, i.e. much less than the 2?10 6 receptors characteristic of A431 cells. Figure S8 shows that EGFR on Hela cells was activated by magnetization of bound MS, demonstrating that activation by local clustering is a phenomenon that can be generalized to other cell types with low receptor densities.
A similar clustering and activation of the EGFR was induced by strv-SPIONs targeted to the chimeric EGFR with the biotinylated ACP tag ( Figure 6). We conclude that higher-order clustering of the receptor is sufficient for activation, i.e. does not require a unique orientation of the extracellular domain or binding of a large antibody molecule.

Discussion
Localized activation of EGFR induced by EGF-coated magnetic microspheres (,1 mm) has been shown to occur exclusively at the attachment loci on the cell membrane [42,43,44]. In the current study, we explored the applicability of much smaller superparamagnetic NPs lacking a natural ligand for the receptor and targeted to the EGFR with a monoclonal antibody that actually blocks ligand binding. The aim was to assess whether local clustering of the receptor by itself leads to activation and downstream signaling.
We demonstrated that simple bivalent coupling by streptavidin of the biotinylated antibody or targeting SPIONs to the cell membrane did not result in EGFR activation. However, exposure of the targeted SPIONs to a strong magnetic field resulted in a time-dependent and ligand-independent EGFR activation. MS consist of individual magnetic domains; a strong magnetic dipolar interaction is induced by the external magnetic field, which propagates and causes clustering of closely placed particles [45]. Exposure of cells to a magnetic field in the presence of nontargeted SPIONs failed to activate the EGFR.
Cells respond to mechanical stimuli through integrin signaling involving focal adhesion kinases (FAK) [46]. However, the forces required for conventional cellular mechanotransduction through integrins are on the order of pN to nN [46,47,48]. Surprisingly, the magnitude of forces measured in this study for activation of EGFR was estimated to be in the sub-pN range, (Supporting Information S1), i.e. considerably less than the force required for integrin mediated mechanotransduction.
Recently Rauch et al. [49] assayed the ability of dextran-coated SPIONs with various surface charges to activate ERK, AKT and EGFR in breast and colon cancer cells either expressing a metastatic RAS mutation or without this mutation. They found strong activation of the RAS pathway in the mutated cells in the absence of a magnetic field. In the absence of a magnetic field the epidermoid carcinoma cell line A431 did not show uptake or activation of our targeted SPIONs after incubations of 15 min at 37uC ( Figure S5.) However, a slow uptake of the SPIONs, presumably by macropinocytosis, occurred upon prolonged incubation. We did not investigate cells with RAS mutations and thus cannot comment about whether non-dextran coated SPIONs might also effect a similar response as that seen by Rauch et al in Another publication featured gold-coated SPIONs coated with a saturating concentration of C225 monoclonal anti-EGFR SPIONs to study the downregulation of the EGFR after prolonged incubation (72 hours) in the absence of an external field whereby the cells autophagosized the particles [50]. C225 was originally characterized for its ability to block ligand binding to the EGFR, to arrest cell proliferation and to promote tumor killing [28,37]. Coupling of the antibody to the NPs resulted in a two-fold increase in cell killing.
The absolute magnetization of MS depends on the magnetization distance and the strength of the applied magnetic field. We achieved full activation of the EGFR on A431 cells with an application of the magnetic field for 3 min at the employed field strength. The same downstream signaling cascade was promoted by magnetic activation with the SPIONs both using antibody targeting or direct binding of streptavidin SPIONs to the biotinylated chimeric EGFR.
The requirement for receptor-receptor interaction in the activation of signal transduction is a signature event of many receptor tyrosine kinases [52] and T cell receptors [53]. Cells with both high and low expression levels of EGFR (A431cells with 2?10 6 compared to HeLa cells with 5?10 4 ) were activated by SPIONs, leading to the conclusion that high local density (in the absence of the natural ligand) suffices for activation of the EGFR and its downstream signaling cascades. Thus, the data reinforce the hypothesis that upregulation of EGFR, interactions with other cell membrane components or sequestration in particular membrane microdomains leading to functional receptor clusters can potentiate cellular growth and migration of cells.

Conclusions
We have demonstrated that physiologically important signaling receptors can be mechanically activated using magnetic nanotechnology. SPION technology constitutes a valuable tool for manipulating cellular processes in order to study and/or modulate functional states. In some human cancers receptor tyrosine kinases such as EGFR are in a constitutively active form due to mutations that abrogate the quiescent, non-aggregated state of the receptor. Our results provide direct evidence for a possible additional mechanism, i.e clustering per se, for the ligand free activation of EGFR in cancer cells expressing high levels of the wild-type receptor. Specific targeting of magnetized superparamagnetic NPs to the EGFR actuates the normal signaling cascade via the induction of a mechanical force of very low magnitude (sub-pN). It follows that magnetic switches used in conjunction with an external magnetic field may be used to control cellular functions in vivo. In particular, it may be possible to specify the location and strength of a signaling mechanism with magnetic control devices or to specifically deliver SPIONs into cells by such techniques. EGFR and other ErbB family members are upregulated in approximately 50% of all human tumors [51]. A prominent example is glioblastoma. Kantelhardt et al. [54] demonstrated that glioma cells expressing highly upregulated EGFR, in contrast to normal brain tissue lacking this receptor, were selectively labeled by quantum dots specifically targeted to the EGFR. SPIONs can be coupled to chemotherapeutic drugs, and animal studies show that uptake into tumors can be promoted by focused magnetic fields [55]. Thus, MS of the type described in this report may constitute useful vehicles for specifically targeting chemotherapeutic drugs to residual tumor cells after surgical resection.

Synthesis of Superparamagnetic Iron Oxide Nanoparticles and Surface Modification with Streptavidin
Superparamagnetic iron oxide NPs (SPIONs) were synthesized by alkaline coprecipitation of magnetite [56] from a mixture of Fe 3+ and Fe 2+ salts (FeCl 3 and FeCl 2 in 2:1 molar ratio) followed by addition of concentrated NH 4 OH solution (25% w/v). The size of fractionated SPIONs was monitored by the transmission electron microscope (Jeol) operated at 120 kV and by dynamic light scattering (Malvern instruments) ( Figure S1 a-d and e). SPIONS of 10-15 nm diameter were covalently coupled to streptavidin such that most particles had one or fewer streptavidin molecules (see Supporting Information S1 for details). Purified strv-SPIONs were resuspended in 20 mM Na-PO 4 , pH 7.6, buffer and further used for synthesis of the magnetic switches (MS).

Preparation of Magnetic Switches
Magnetic switches (MS) capable of targeting EGFR were prepared (except when cells expressed ACP-EGFR, see below) by incubating a 5-fold molar excess of strv-SPIONs, in terms of biotin binding capacity, with biotinylated anti-EGFR monoclonal antibody 528 (Dianova) in 20 mM Na-PO 4 buffer (pH 7.4) at 25uC for 15 min. Finally, the MS were incubated with a two-fold molar excess of biocytin-Alexa fluorophore (Invitrogen) (in terms of biotin binding capacity of strv-SPION) for 15 min and washed by magnetic separation several times to remove unbound fluorophore.

Covalent Labeling of Chimeric ACP-EGFR with Biotin-CoA on Expressing Cells
Biotin-CoA was synthesized and purified by HPLC C 18 chromatography as described in the literature [57]. We constructed a chimeric ErbB1 with an acyl carrier protein tag [58] between the signal peptide and the ErbB1 mature protein sequence. The serine that accepts a CoA derivative is located at position 36 in the sequence of the mature expressed protein. The 78 amino acid DNA sequence for the ACP tag was extracted from the AGAplasmid (kind gift of N. Johnsson) by PCR and inserted between the signal sequence and the amino terminus by creating a new Nhe1 restriction enzyme cleavage site using site-directed mutagenesis of wild-type ErbB1 in pcDNA3 or a modified pcDNA3.1zeo 2 plasmid. Stably-transfected Hela cell lines were selected by antibiotic resistance and single cell cloning. The chimeric EGFR behaves as wildtype EGFR as reported elsewhere [59].
Cells expressing the chimeric ACP-EGFR were labeled covalently with 1 mM CoA-biotin by enzymatic reaction at room temperature for 20 min with E. coli PPTase (phosphopantetheinyl transferase) [57,60] which links biotin through a phosphodiester bond to the serine hydroxyl in the ACP tag. For experiments described in the text strv-SPION (10 mg/ml) NPs were added to the cells at 15uC for 15 min and excess SPIONs removed by washing before exposure of the cells to a magnetic field ( Figure 6). Ligand-Free Activation of EGFR Signaling PLOS ONE | www.plosone.org

Cell Culture
The following cell lines were used: A431, Human epidermoid carcinoma (ATCC CRL 1555) Clone E3 from E. Helmreich, Würzburg; Hela SS6; Hela stably expressing actin-GFP; and Hela stably expressing EGFR with an amino terminal ACP tag that can be covalently labeled by phosphopantetheinyl transferase with CoA-biotin. Cells were seeded on 24 mm square coverslips in complete DMEM and grown to 70% confluency. Cells were starved overnight and washed 3 times at 1 hr intervals prior to use in order to achieve a quiescent state and thus minimize the serumor autocrine-induced activation of EGFR.

Binding to Cells and Magnetization of MS
Cells were incubated in Tyrode's buffer containing 0.1% BSA and 20 mM glucose at 15uC for 15 min with MS (5 mg/ml < 6 nM Mab) and excess, unbound MS were removed by 3 washes with the same buffer. MS were magnetized by placing 4 permanent Neodymium magnets in a quadrapole configuration on the 4 sides of the glass coverslip at room temperture. After the allotted magnetization time coverslips were removed from the magnetic field and immediately fixed with 3.7% paraformaldehyde for 30 min on ice (unless otherwise indicated in the text) and then stained for activated EGFR or other cellular proteins as indicated in the figure legends.

Detection of Phosphorylation of EGFR (pY-EGFR) by Immunofluorescence
Cells were fixed in 3.7% PFA for 30 min at 4C, permeabilized and blocked with 0.1% Tween 20, 1% BSA in PBS at room temperature for 1 hr followed by incubation with rabbit antiphosphotyrosine EGFR (pY-EGFR 1068 or pY-EGFR 1148 (Cell Signaling Technology) at 100-300 ng/ml. Primary antibodies were detected using secondary F(ab) 2 fragments of goat anti-rabbit IgG (GARIG) labeled with Cy3 or Cy5 (Jackson Labs) at 1 mg/ml. Coverslips were mounted on microscope slides in Mowiol (Polysciences).

Western Blot Analysis
For western blot analysis, A431 cells were grown in 24 well plates and treated with MS in the presence or absence of a magnetic field. Total cell lysates were obtained with Phosphosafe Reagent (Novagen) and equal amounts of protein were separated by a polyacrylamide gel electrophoresis, blotted to PVDF membranes and probed with rabbit primary antibody (pY-EGFR 1148, Cell Signaling Technology) and HRP-conjugated goat antirabbit IgG. Signal was developed by chemiluminescent (Supersignal West Pico, Pierce) and imaged with Kodak Biomax Xray film.

Detection of MS in Early Endosomes
Magnetized cells were fixed after an additional incubation for 20 min at 37uC after removal from the magnetic field. Early endosomes were visualized by indirect immunofluorescence using 250 ng/ml rabbit polyclonal antibody (Abcam) against early endosomal antigen (EEA1) and goat anti-rabbit-Cy3 antibody (Jackson Labs).

Shc Activation and Recruitment Assay
After binding of MS and magnetization, A431 cells were incubated for 10 min at 37 uC before fixation. Cells were stained with rabbit anti-phosphotyrosine 317 Shc antibody (Cell Signaling Technology) and GARIG-Cy5 (Jackson Labs). The distributions of Alexa 488 MS signal and Shc signal were detected by GARIG-Cy5. In the parallel control experiment the same incubation conditions were used without the application of a magnetic field.

Confocal Microscopy
Confocal laser scanning fluorescence microscopy was carried out with an LSM 510 Meta system (Carl Zeiss, Jena) using a 639, 1.4 NA Plan-Apochromat oil immersion objective. MS (Alexa 488) were excited at 488 nm with the Argon ion laser and detected with a 520/30 nm bandpass filter. GARIG-CY3 was excited with a 532 nm DSSP laser and its emission detected .585 nm. GARIG-Cy5 was excited with a 633 nm HeNe laser and its emission detected .650 nm. Images were recorded with 4 fold averaging in the XY dimension with 2 s sampling time per frame. All fluorescence images were recorded in multichannel mode in order to avoid the crosstalk between channels. Transmission images were acquired in DIC mode with either the 488 nm or the 532 nm laser line.

Image Processing
Antibody and MS fluorescence signals were quantified using thresholding to segment images after background subtraction using the public domain NIH image J programs (available on the Internet at http://rsb.info.nih.gov/nih-image). Colocalization analyses of confocal stacks of 0.3 m-separated images were performed after deconvolution with Huygens image processing software (Scientific Volume Imaging, Netherlands) and twodimensional histograms representing the distributions plotted from these data. Mander's coefficients were determined from deconvolved stacks by an Image J plugin.

Scanning Electron Microscopy
SEM imaging of MS targeted to EGFR on A431 cell surfaces was carried out with a Hitachi S 5500 system operated at 30 kV. Cells were grown on 1 cm 2 Si wafers treated with collagen, incubated with MS in the presence or absence of an external magnetic field and fixed with 2.5% EM grade glutaraldehyde followed by dehydration steps through a 30-100% ethanol gradient. In view of the low electron density and insulating nature of iron oxide, cells were treated with tannic acid and uranyl acetate for conductivity and contrast enhancement. Images were acquired over various magnifications in scanning mode with a field emission electron gun. Figure S1 TEM images of as synthesized SPIONs before (a,c) and after (b,d) magnetic fractionation. SPIONs with fairly uniform size and shape were obtained after magnetic fractionation. (e) hydrodynamic size distribution from dynamic light scattering of magnetically fractionated SPIONs with the mean diameter of , 15 nm.