EGFRvIII Antibody–Conjugated Iron Oxide Nanoparticles for Magnetic Resonance Imaging–Guided Convection-Enhanced Delivery and Targeted Therapy of Glioblastoma Free

Despite the use of conventional therapeutic modalities such as surgery, chemotherapy, and ionizing radiation, the prognosis in patients with malignant gliomas remains poor (1). Virtually all glioblastoma multiforme (GBM) tumors, the most common malignant glioma, recur at the site of their initial treatment due to the presence of infiltrating cancer cells in the surrounding normal brain that resist therapy or go untreated. Infiltrating cancer cells include a subpopulation of glioblastoma stem cells (GSC) shown to be integral to tumor development, perpetuation, and therapy resistance (2, 3). Imaging and targeted therapy of infiltrating GBM cells within the normal brain remain limited.

The epidermal growth factor receptor variant III (EGFRvIII) is a tumor-specific mutation that is expressed in malignant gliomas and not in the normal brain. This mutation encodes a constitutively active tyrosine kinase that enhances tumorigenicity and accounts for radiation and chemotherapy resistance (4, 5). The 801-bp in-frame deletion in the extracellular domain of the EGFR results in the fusion of normally distant EGFR gene and protein sequences (6, 7). The 14-amino-acid fusion junction sequence has been chemically synthesized and used to create an anti–synthetic peptide antibody that is highly specific for the deletion mutant EGFR protein compared with the intact EGFR protein (8, 9). Vaccination of the fusion junction peptide sequence has been shown to be efficacious immunotherapy in syngeneic murine models and in humans with two consecutive and one multi-institutional phase II trials (10).

The magnetic nanoparticle has emerged as a potential multifunctional clinical tool that can provide cancer cell detection by magnetic resonance imaging (MRI) contrast enhancement as well as therapy by cancer cell–targeted delivery of therapeutic agents (antibodies, drugs, and small-molecule inhibitors) or by local hyperthermia generated by absorbing energy from an alternating magnetic field. Iron oxide nanoparticles (IONP) in the size range of 10 to 25 nm have unique magnetic properties, generating significant transverse T₂ relaxation time shortening and susceptibility effects resulting in strong T₂-weighted contrast on MRI (11). IONPs can evade the immune system and target cancer cells for destruction while simultaneously providing MRI contrast. Most IONPs are biodegradable and considered to have low toxicity (12). IONPs have been used in the clinical setting with humans (13). Currently, various formulations of IONPs have been developed for drug delivery schemes (14), magnetic cell separation and cell targeting (15), magnetic resonance imaging (MRI) contrast enhancement (16–20), and hyperthermia treatment of cancer (17, 21–23).

Cell-specific imaging by nanotechnology for detection and treatment monitoring holds great promise for the therapy of various cancers including central nervous system (CNS) tumors (24, 25). However, a major barrier in the use of nanotechnology for brain tumor applications is the difficulty in delivering nanoparticles to intracranial tumors. Conventional systemic delivery is limited due to the nonspecific nanoparticle uptake by the reticuloendothelial system and problems in penetrating the blood-brain barrier (BBB). Convection-enhanced delivery (CED) is a minimally invasive surgical procedure that provides fluid convection in the brain by a pressure gradient that bypasses the BBB. Therapeutic agents can be delivered into the brain by CED in high concentrations (26, 27) without toxicity to normal tissue and organs commonly associated with systemic delivery. The use of CED can also allow for therapeutic targeting of infiltrating cancer cells, a major cause for brain tumor recurrence after surgery. We report the use of IONPs conjugated to an anti–synthetic peptide antibody (EGFRvIIIAb) specific to the deletion-mutant EGFR for image-guided CED in a mouse glioma model. The EGFRvIIIAb-IONP complex can provide MRI contrast enhancement of human glioblastoma cells in vitro and an antitumor effect both in vitro and in vivo after CED.

The human glioblastoma cell line U87MG was obtained from the American Type Culture Collection within the last 6 months and maintained in standard culture conditions. The U87MG cell line was stably transfected with either a plasmid for overexpression of the deletion mutant EGFRvIII (U87ΔEGFRvIII) or the wild-type (wt) EGFR (U87wtEGFR) and tested by Western blot analysis (Supplementary Fig. S1). Neurospheres were harvested from patient GBM specimens (patients #74, 1002, and 30) at the time of surgical resection and cultured using serum-free medium supplemented with growth factors. Patient tumor specimens were harvested with approval by the Emory University Institutional Review Board (protocol #642-2005). Glioblastoma neurospheres were tested by Western blot analysis (Supplementary Fig. S2) for the GSC stem cell marker CD133 (Cell Signaling), EGFRvIII (GenScript Corp.), and wt EGFR (Cell Signaling). Overexpression of the deletion mutant EGFRvIII confers enhanced tumorigenicity in immunocompromised rodents (4). Six- to seven-week-old athymic nude (nu/nu) mice were used, and all procedures were done with approval by the Institutional Animal Care and Use Committee of Emory University. The IgG polyclonal EGFRvIII antibody (6 nm and 150 kDa) was generated by GenScript in rabbits and purified as described in a prior publication (8). Briefly, the rabbit polyclonal EGFRvIIIAb represents an anti–synthetic peptide antibody that reacts to the fusion junction of the deletion mutant EGFRvIII receptor expressed in human glioblastoma tumors.

Briefly, activation of the carboxyl groups on the IONPs was performed for conjugation of the EGFRvIII antibody, after addition of an activation buffer, ethyl dimethylaminopropyl carbodiimide (EDC), and sulfo-NHS. The EDC/NHS solution was mixed vigorously with the IONPs at 25°C for 15 minutes. Excess EDC and sulfo-NHS were removed from the activated nanoparticles by three rounds of centrifugation (1,000 × g) and resuspension in PBS using Nanosep 10K MWCO OMEGA membrane (Pall Life Sciences). The IONPs with activated carboxyl groups were then reacted with the EGFRvIII antibody (50 μL at 2 mg/mL) at 25°C for 2 hours, and the reaction mixture was stored at 4°C overnight. Excess antibody was removed by three rounds of centrifugation and resuspension in PBS using 300K MWCO OMEGA membranes. Mobility shift in 1% agarose gel was visualized by staining with 0.25% Coomassie brilliant blue in 45% methanol, 10% acetic acid for 1 hours and destaining in 30% methanol, 10% acetic acid overnight (Supplementary Fig. S3).

Glioblastoma cells (U87ΔEGFRvIII) were seeded in triplicate in 60-mm flat-bottomed plates and incubated overnight at 37°C. Confluent monolayers of cells (1 × 10⁶) were incubated with the IONP or EGFRvIIIAb-IONP solution (0.15 mg/mL) for 1 or 2 hours for MRI experiments. Washing of cells was performed with 10% PBS to remove the excess nanoparticle solution. Treated cells were collected after scraping and placed in 10-mL tubes containing warm 0.8% agarose solution. MRI measurements were made with a 3-T clinical capable scanner. The T₂ relaxation times of cells treated with IONPs were measured using a multi-echo fast spin echo sequence with 32 TE values ranging from 6 to 180 ms and an interval of 6 ms. The T₂ relaxation time was calculated by fitting the decay curve using the nonlinear monoexponential algorithm of M_(TE) = M₀ × exp(−TE/T₂). See Supplementary Methods for transmission electron microscopy (TEM) studies.

Human astrocytes were kindly provided by the Yong Laboratory at the University of Calgary, Alberta, Canada. The astrocytes were cultured based on a prior published protocol (28). Toxicity experiments were performed on human astrocytes seeded in triplicate in 48-well flat-bottomed plates (10⁵ astrocytes per well). Cells were treated with free IONPs (0.3 mg/mL) or control vehicle (serum-free medium) for 1 hour at 37°C. Glioblastoma cells (U87MG and U87ΔEGFRvIII) were seeded in triplicate in 48-well flat-bottomed plates (80,000 cells per well) and incubated overnight at 37°C. Confluent monolayers of cells were washed with PBS and then incubated with control (PBS), IONPs (0.2 mg/mL), EGFRvIIIAb (0.2 mg/mL), or EGFRvIIIAb-IONPs (0.2 mg/mL) for 1 hour at 37°C. Cells were then washed with PBS and a crystal violet cell viability assay was performed at 0, 1, and 3 days. Cells were stained with 100 μL of crystal violet solution (1% crystal violet, 1% HCl, and 10% ethanol) at 37°C for 30 minutes. Absorbance measurements were performed on a microtiter plate reader at a wavelength of 570 nm. Absorbance values are presented as the mean of three wells per treatment ± SD.

Glioblastoma cells (U87MG, U87ΔEGFRvIII, and U87wtEGFR) and neurospheres from patients #30 and 74 were seeded in six-well flat-bottomed plates (500,000 cells per well) and incubated overnight at 37°C. Cells were treated with control IgG (0.3 mg/mL), IONPs (0.3 mg/mL), EGFRvIIIAb (0.3 mg/mL), and EGFRvIIIAb-IONPs (0.3 mg/mL) for 2 to 3 hours in PBS at 37°C. Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer [50 mmol/L Tris (pH 8.0), 150 mmol/L NaCl, 5 mmol/L EDTA, and 1% NP40, with protease and phosphatase inhibitors]. Western blot analysis of protein levels was performed with cleaved caspase-3 and caspase-3 primary antibodies (Cell Signaling) according to the manufacturer's recommendations.

For EGFR signaling studies, glioblastoma cells were treated with control (PBS), IONPs (0.5 mg/mL), EGFRvIIIAb (0.5 mg/mL), and EGFRvIIIAb-IONPs (0.5 mg/mL) for 2 hours in PBS at room temperature. Cell lysates were harvested in RIPA buffer [50 mmol/L Tris (pH 8.0), 150 mmol/L NaCl, 5 mmol/L EDTA, and 1% NP40, with protease and phosphatase inhibitors]. Western blot analysis of protein levels was performed with primary antibodies against phospho-EGFR (Transduction Laboratories), phosho-Akt (Cell Signaling), Akt (Cell Signaling), phospho-extracellular signal–regulated kinase (ERK)-44/42 (Cell Signaling), and ERK44/42 (Cell Signaling) according to the manufacturer's recommendations. Horseradish peroxidase (HRP)–conjugated secondary antibodies (Dako) and a chemoluminescent HRP detection solution (Denville Scientific, Inc.) were used to detect the proteins.

Anesthetized athymic nude mice were placed in the stereotaxic instrument and U87ΔEGFRvIII cells (5 × 10⁵) were stereotactically inoculated into the right striatum, 3 mm below the dural surface, on day 0. On day 7 after tumor implantation, mice were randomized into four groups: (a) CED of HBSS (untreated control), (b) CED of IONPs, (c) CED of EGFRvIIIAb, and (d) CED of EGFRvIIIAb-IONPs. All animals underwent CED of a 10-μL volume at a rate of 0.5 μL/min (20 minutes of total infusion) on day 7 using the same coordinates as used earlier for tumor cell implantation. For the CED methods, see Supplementary Methods.

Mice inoculated with U87ΔEGFRvIII cells (5 × 10⁵) were also followed by MRI to investigate the initial volume of distribution (V_D) and volume of dispersion (V_DI) of IONPs over a period of 11 days after CED. Animals underwent CED of IONPs (n = 3 free IONPs and n = 3 EGFRvIIIAb-IONPs) as described above on day 9 after tumor implantation. These animals were scanned on a 4.7-T animal MRI scanner using a dedicated mouse coil (Varian Unity) on days 0, 4, 7, and 11 after CED. The V_D was determined for each animal (n = 4) on days 0, 4, 7, and 11 after CED. The V_DI of the IONPs on days 4, 7, and 11 was determined by subtracting the V_D at day 0 from the V_D calculations on days 4, 7, and 11 after CED of the IONPs (e.g., V_DI day 4 = V_D day 4 − V_D day 0). For V_D and V_DI measurements, see Supplementary Methods.

Mouse brains were harvested and fixed with 10% neutral buffered formalin on days 7 and 9 after tumor implantation. Coronal sections were made at the level of the needle tract to mark the center of the xenograft tumor. Serial sections of the cerebral hemisphere were examined from each animal. Tissue blocks were embedded in paraffin and sectioned (5 μm). Prussian blue staining was performed after slides of formalin-fixed tissue were placed in a coplin jar containing a 1:1 mixture of 5% potassium ferrocyanide and 5% HCl for 30 minutes at 37°C in a water bath. The slides were rinsed well with distilled water and counterstained with nuclear fast red for 20 minutes (Supplementary Fig. S4). EGFRvIII immunohistochemistry was performed (day 7 after tumor implantation) after deparaffinization of tissue sections (Supplementary Fig. S5). The primary rabbit polyclonal EGFRvIII antibody (GenScript) and a biotinylated antirabbit secondary antibody provided in a rabbit ABC staining system (Santa Cruz Biotechnology) were used.

All athymic nude mice were observed daily to monitor external appearance, feeding behavior, and locomotion. Animals were sacrificed at the first sign of an adverse event (paresis, inability to feed) and brains were removed.

Cell viability data represent the average of three independent absorbance values on day 3 of the crystal violet assay. Statistical cell viability differences were assessed using unpaired two-sample Student's t test assuming equal variance. Animal survival data were entered into Kaplan-Meier plots and statistical analysis was performed using log-rank test. P = 0.05 was considered statistically significant.

Amphiphilic triblock copolymer–coated IONPs (10 nm in size; provided by Ocean Nanotech, LLC) were covalently conjugated to a purified rabbit polyclonal EGFRvIIIAb (Supplementary Fig. S1). The IONP surface coating provides a stable hydrophobic protective inner layer around a single crystal of IONP with carboxylate groups in the outer layer for functionalization (29). Bioconjugation was performed with the amino-terminal fragment of the antibody and carboxyl groups on the copolymer coating of the IONPs (Fig. 1). Confirmation of the antibody conjugation to IONPs was performed by agarose gel electrophoresis analysis (Supplementary Fig. S3).

The binding affinity of EGFRvIIIAb-IONPs to human GBM cells that overexpress the EGFRvIII mutated protein was evaluated. The EGFRvIIIAb-IONP conjugate was able to selectively bind to U87ΔEGFRvIII cells and induce T₂-weighted MRI contrast enhancement after 1 and 2 hours of treatment (Fig. 2; Table 1), as evidenced by the MRI signal drop compared with the control samples of cells treated with free IONPs. Such strong T₂ shortening and the magnetic susceptibility effects of IONPs lead to spin dephasing and substantial MRI signal drop, which generated a “darkening” contrast as seen in the T₂-weighted MR images. TEM confirmed binding of EGFRvIIIAb-IONPs to the cell surface of EGFRvIII-expressing glioblastoma cells as shown in Supplementary Fig. S6. No evidence of endocytosis and endosome-filled EGFRvIIIAb-IONPs was found in glioblastoma cells based on TEM. In contrast, treatment of cells with free IONPs revealed nonspecific cell surface binding, uptake, and IONP-filled endosomes within the cytoplasm of U87ΔEGFRvIII cells (Supplementary Fig. S7).

To determine whether IONPs are toxic to human astrocytes, cells were treated with control vehicle (serum-free medium) or IONPs (0.3 mg/mL) for 1 hour. After cell washing, a crystal violet cell viability assay was performed. There was no significant toxicity found with human astrocytes 3 days after treatment with IONPs (P < 0.001; Fig. 3A).

Human glioblastoma cells (U87-MG or U87ΔEGFRvIII) were incubated with control (PBS), EGFRvIIIAb, IONPs, or EGFRvIIIAb-IONPs for 1 hour (Fig. 3B and C). Cells were washed and cell viability assay was performed at 0, 1, and 3 days after treatment. At day 3, there was a significant decrease in cell survival in U87ΔEGFRvIII cells treated with IONPs, EGFRvIIIAb, or EGFRvIIIAb-IONPs (P < 0.001). There was a less pronounced but statistically significant decrease in cell survival in U87MG cells treated with IONPs, EGFRvIIIAb, or EGFRvIIIAb-IONPs (P < 0.01). Nonspecific uptake of free IONPs likely accounts for cell toxicity and the antitumor effect found in both glioblastoma cell lines (Supplementary Fig. S7).

Human glioblastoma cells (U87MG, U87ΔEGFRvIII, and U87wtEGFR) and neurospheres were incubated with control IgG, EGFRvIIIAb, IONPs, and EGFRvIIIAb-IONPs (Fig. 4). Western blot analysis revealed elevated levels of cleaved caspase-3 in glioblastoma cells (Fig. 4A) and neurospheres (Fig. 4B) treated with EGFRvIIIAb-IONPs. No levels of cleaved caspase-3 were found with treatment of glioblastoma cells with free IONPs, control IgG, or EGFRvIIIAb. Elevated levels of cleaved caspase-3 and apoptosis were found in human glioblastoma neurospheres harvested from patients #30 and 74 after treatment with the EGFRvIIIAb-IONPs (Fig. 4B). Glioblastoma neurospheres from patient #30 are characterized by EGFRvIII, wt EGFR, and CD133 GSC marker expression, whereas neurospheres from patient #74 express wt EGFR (Supplementary Fig. S2).

Ligand activation of the EGFR leads to a series of downstream signaling events after phosphorylation and activation (30, 31). Western blot analysis of EGFR cellular signaling proteins was performed in glioblastoma cells (U87MG, U87wtEGFR, and U87ΔEGFRvIII) treated with control (PBS), IONPs, EGFRvIIIAb, or EGFRvIIIAb-IONPs (Fig. 4C). Levels of phosphorylated EGFR, phospho-Akt, Akt, ERK44/42, and phospho-ERK44/42 were determined. Both U87wtEGFR and U87ΔEGFRvIII showed elevated levels of phosphorylated EGFR as expected due to the overexpression of EGFR in these cell lines. U87ΔEGFRvIII cells showed the greatest EGFR phosphorylation due to the constitutive activity of the EGFRvIII receptor. Treatment of these cells with EGFRvIIIAb-IONPs showed a lower level of EGFR phosphorylation. Treatment of U87wtEGFR and U87ΔEGFRvIII glioblastoma cells with either IONPs or EGFRvIIIAb-IONPs resulted in less phosphorylation of Akt and no phosphorylation of ERK, suggesting that IONPs may affect downstream EGFR signaling in glioblastoma cells. Treatment of U87wtEGFR and U87ΔEGFRvIII glioblastoma cells with EGFRvIIIAb alone resulted in increase in phosphorylated Akt and ERK levels.

Athymic nude mice (n = 6) implanted with human GBM xenografts that express the mutated EGFRvIII protein (U87ΔEGFRvIII) underwent CED of EGFRvIIIAb-IONPs (n = 3) or IONPs (n = 3) in aqueous solution (0.2 mg/mL concentration; 10 μL volume; 0.5 μL/min rate for a total of 20 minutes) 9 days after tumor implantation. MRI was performed before CED to confirm proper growth of the xenograft in each mouse brain (Fig. 5) and then 0, 4, 7, and 11 days after CED of the nanoparticles to determine localization, initial V_D, and V_DI of the IONPs administered (Fig. 5). Average V_D values were 11.2, 19.2, 22.7, and 22.9 mm³ on days 0, 4, 7, and 11, respectively. The averaged V_DI values were 8.0, 11.5, 15.5 mm³ on days 4, 7, and 11 after CED. There was no statistically significant difference in the V_D and V_DI of free IONPs and EGFRvIIIAb-IONPs. Prussian blue staining was performed to confirm the presence of EGFRvIIIAb-IONPs in human xenograft tumors and the surrounding mouse brain after CED (Supplementary Fig. S4).

Several initial findings support the feasibility of the MRI-guided CED of IONPs in human xenograft brain tumors: (a) EGFRvIIIAb-IONPs can be distributed within or adjacent to brain tumors with CED; (b) outstanding contrast induced by IONPs enables MRI to monitor and follow the distribution of the nanoparticle complex in the tumor and in the surrounding brain after CED; (c) intracerebral infusion is safe, as all animals survived from the procedures and showed no signs of toxicity; (d) CED led to a long retention of the EGFRvIIIAb-conjugated nanoparticles and slow dispersion of the agents at the tumor site as shown on MRI; (e) MRI can delineate and quantify the extent of dispersion of the IONP complex after CED with T₂-weighted MRI.

Animal survival studies were performed in four groups of mice implanted with highly tumorigenic U87ΔEGFRvIII xenografts (Fig. 6; Supplementary Fig. S5). Each group of animals (n = 10) underwent CED of control (HBSS), unconjugated IONPs (0.2 mg/mL), EGFRvIIIAb (0.2 mg/mL), or EGFRvIIIAb-IONPs (0.2 mg/mL) 7 days after tumor implantation. All the animals that underwent HBSS CED were dead by 14 days after tumor implantation (median survival, 11 days). A significant increase in survival was found in animals that underwent CED of IONPs (median survival, 16 days), EGFRvIIIAb (median survival, 17 days), and EGFRvIIIAb-IONPs (median survival, 19 days) compared with animals that underwent CED of HBSS (P < 0.001). Both the EGFRvIIIAb and EGFRvIIIAb-IONP treatment groups have a single survivor after 120 days of survival studies.

Accurate targeting and better delivery efficiency are two major goals in the development of therapeutic agents for the treatment of malignant brain tumors. Ideally, a therapeutic agent would be able to overcome the BBB and selectively be enriched in tumors with minimal toxicity to normal tissues. Conjugation of agents to antibodies or other ligands that bind to antigens or receptors that are usually abundantly or uniquely expressed on the tumor surface represents a promising approach in the treatment of glioblastoma tumors (32, 33). Unfortunately, the ability to image therapeutic agents in the brain for adequate tumor delivery and the proper assessment of treatment efficacy remain limited.

The use of nanotechnology is now being applied to CNS cancer applications (13, 25, 34–39). One potential problem that can attenuate the targeted therapy and imaging of CNS tumors by systemic delivery of nanoparticles is their being “trapped” in the liver, spleen, and circulating macrophages after i.v. administration due to nonspecific uptake. Additionally, systemic delivery is also limited by the BBB, nontargeted distribution, and systemic toxicity. CED is an approach developed to overcome the obstacles associated with current CNS agent delivery (26, 27, 40) and is increasingly used to distribute therapeutic agents for treatment of malignant gliomas (27). Recently, multiple clinical trials have been performed using CED for the treatment of recurrent GBM (32, 33, 41). In CED, a small hydrostatic pressure differential imposed by a syringe pump to distribute infusate directly to small or large regions of the CNS is used in a safe, reliable, targeted, and homogeneous manner (42). Difficulty in CED imaging of therapeutic agents in the brain is one major limitation of this approach. Groups have radiolabeled their therapeutic agent, co-infused their agent with radiolabeled albumin (¹²³I-labeled albumin), or used liposomes containing an MRI contrast agent (e.g., gadoteridol) for CED imaging (33, 43, 44). Radiolabeling of therapeutic agents relies on single-photon emission computerized tomography for agent imaging, which is a low-resolution imaging modality of the brain. Co-infusion of an MRI contrast agent or radiolabeled albumin may not allow for precise therapeutic agent distribution analysis in the brain due to the differences in molecular weight and surface properties between each infusate. Recently, CED of biodegradable, nonfunctionalized maghemite magnetic nanoparticles has been depicted by MRI in a normal rat brain model (45). Direct imaging of magnetic nanoparticles by MRI can permit distribution studies of nanoparticles in the brain after CED.

Currently, two major types of systemic anti-EGFR agents have entered the clinical setting: anti-EGFR antibodies and small-molecule EGFR tyrosine kinase inhibitors (46–48). These agents have shown modest efficacy in patients with GBM tumors. The development of new agents that can target the EGFR deletion mutant EGFRvIII can permit direct targeting of glioblastoma cells while sparing the normal brain (9, 49–52). Furthermore, the targeting of GSCs and/or GSC-specific molecules (e.g., CD133) may form the basis of more effective treatments against glioblastoma tumors and prevention of relapse (2).

We report for the first time the use of IONPs bioconjugated to an EGFRvIII deletion mutant antibody for MRI-assisted CED and targeted therapy of human glioblastoma. The bioconjugated EGFRvIIIAb-IONPs permit MRI contrast enhancement or “darkening” of U87ΔEGFRvIII cells with T₂-weighted MRI. The EGFRvIIIAb-IONPs caused a significant decrease in glioblastoma cell survival, and a greater antitumor effect was found after treatment of EGFRvIII-expressing glioblastoma cells with EGFRvIIIAb-IONPs in comparison with human glioblastoma cells, which did not express the EGFR. This provided further evidence of a biomarker-targeting effect by the EGFRvIIIAb-IONPs. The pronounced antitumor effect of free IONPs in glioblastoma cells is likely due to the nonspecific uptake of the nanoparticles by the tumor cells. Uptake of IONPs by glioma cells has been shown in culture and in vivo in the past (53, 54). The influence of surface functionalization has recently been shown to enhance the internalization of magnetic nanoparticles in cancer cells (55). Our IONPs are functionalized through surface coating of amphiphilic polymers, which may promote uptake within glioblastoma cells and result in cell toxicity.

No significant toxicity was found with IONP treatment in human astrocytes or in animals after intracerebral administration. No toxicity to human astrocytes and a significant killing effect of both free IONPs and EGFRvIIIAb-IONPs form the basis of targeted therapy of GBM cells in the brain. Apoptosis was determined to be a mode of cell death after treatment of glioblastoma cells and neurospheres by the conjugate EGFRvIIIAb-IONP. Apoptosis was found after treatment of GSC-containing neurospheres harvested from a glioblastoma patient (patient #30) with elevated expression of EGFRvIII and the GSC marker CD133. Although we were not able to find apoptosis induction in glioblastoma cells after treatment with free IONPs or the EGFRvIIIAb alone, a significant antitumor effect was found both in vitro and in our animal survival studies. Mechanistic studies suggest that EGFR downstream signaling may be affected by IONPs with less EGFR phosphorylation after glioblastoma cell treatment with EGFRvIIIAb-IONPs. Furthermore, IONPs resulted in less phosphorylation of Akt and ERK in glioblastoma cells overexpressing the EGFR.

CED of IONPs in a mouse glioma model results in MRI contrast of the nanoparticles and effective intratumoral and peritumoral distribution of nanoparticles in the brain. A significant therapeutic effect was found after CED of both IONPs and EGFRvIIIAb-IONPs in mice. Dispersion of the nanoparticles over days, after the infusion has finished, may potentially target infiltrating tumor cells outside the tumor mass that are potentially responsible for tumor recurrence and the demise of patients. Use of bioconjugated magnetic nanoparticles may permit the advancement of CED in the treatment of malignant gliomas due to their sensitive imaging qualities on standard T₂-weighted MRI and therapeutic effects. Better targeting of infiltrative glioblastoma tumors by MRI-guided CED of magnetic nanoparticles may provide more effective treatment of these devastating brain tumors. We recognize that the rodent model may not be the ideal animal model to study CED; however, our studies provide a proof-of-principle concept for future CED studies using other cancer nanotechnology agents.

In conclusion, the preclinical use of IONPs conjugated to a glioblastoma-specific antibody may form the basis of a future clinical trial of image-guided CED of magnetic nanoparticles for patients with glioblastoma tumors.

No potential conflicts of interest were disclosed.

Grant Support: NIH grant NS053454 (C.G. Hadjipanayis), NIH Center for Cancer Nanotechnology in Excellence Program grant 54 CA119338-01 (H. Mao), NIH grant P50CA128301-01A10003 (H. Mao and C.G. Hadjipanayis), EmTech Bio, Inc. (H. Mao), Southeastern Brain Tumor Foundation (C.G. Hadjipanayis), the Georgia Cancer Coalition, Distinguished Cancer Clinicians and Scientists Program (C.G. Hadjipanayis), and the Dana Foundation (C.G. Hadjipanayis).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.