The effect of dose and selection of two different ligands on the deposition and antitumor efficacy of targeted nanoparticles in brain tumors

Abstract

Deposition of nanoparticles to tumors often can be enhanced by targeting receptors overexpressed in a tumor. However, a tumor may exhibit a finite number of a biomarker that is accessible and targetable by nanoparticles, limiting the available landing spots. To explore this, we selected two different biomarkers that effectively home nanoparticles in brain tumors. Specifically, we used either an α_vβ₃ integrin-targeting peptide or a fibronectin-targeting peptide as a ligand on nanoparticles termed RGD-NP and CREKA-NP, respectively. In mouse models of glioblastoma multiforme, we systemically injected the nanoparticles loaded with a cytotoxic drug at different doses ranging from 2-8 mg/kg drug. The upper dose threshold of RGD-NP is ~2 mg/kg. CREKA-NP reached its upper dose threshold at 5 mg/kg. For both targeted nanoparticle variants, higher dose did not ensure higher intratumoral drug levels but it contributed to elevated off-target deposition and potentially greater toxicity. A cocktail combining RGD-NP and CREKA-NP was then administered at a dose corresponding to the upper dose threshold for each formulation resulting in a 3-fold higher intratumoral deposition than the individual formulations. The combination of the two different targeting schemes at the appropriate dose for each nanoparticle variant facilitated remarkable increase in intratumoral drug levels that was not achievable by a sole targeting nanoparticle alone.

TOC/Abstract graphic:

A tumor may present limited ‘landing spots’ that are accessible and targetable by nanoparticles. To explore this, we employed two different peptides that effectively direct nanoparticles in brain tumors. We prepared two variants of the same drug-loaded nanoparticle targeting either αvβ3 integrin or fibronectin. Using different doses within the commonly used range of nanoparticles for chemotherapy (2-8 mg/kg drug), each targeted nanoparticle variant reached its upper threshold at a different dose. Injecting the αvβ3 integrin-targeting and fibronectin-targeting nanoparticles together, the intratumoral level of the drug was significantly increased by combining the optimal dose for each nanoparticle variant.

INTRODUCTION

Various strategies have been used to target nanoparticles to upregulated receptors in the tumor microenvironment. Over the past two decades, these efforts have generated a wealth of information on different types of ligands.^1–15 While the dosage of such nanoparticle systems has been selected primarily by considering the maximum tolerance dose of the nanoparticle’s drug cargo and the toxicity of the nanomaterial, it is also important to identify the upper dosage threshold of a targeted nanoparticle beyond which there is no additional benefit in terms of increase in intratumoral deposition of the nanoparticle. Following systemic administration, the blood-circulating targeted nanoparticles arrive at the tumor site and start consuming the cancer-related receptor, which gradually decrease the binding chances of forthcoming particles due to limited availability of ‘landing spots’ often caused by transient receptor saturation.^{16, 17} It is important to emphasize that the intratumoral deposition of targeted nanoparticles is a complex process and depends on various factors including the effect of ligand type on blood circulation, overall immunogenicity of the ligand, spatiotemporal expression of the selected receptor, ligand density on each nanoparticle. The purpose of this work was not to mechanistically assess the influence of these factors on the performance of two different targeting systems. Using two commonly used targeting ligands, our goal was to 1) identify the highest dose for each targeting variant that beyond that dose the intratumoral deposition is plateaued, and 2) use both targeted nanoparticle variants together at their corresponding highest effective dose and evaluate whether the intratumoral deposition can exceed the plateaus of each individual formulation.

In this study, we used mesoporous silica nanoparticles (MSNs), which can conveniently accommodate a high drug cargo and targeting ligands on their surface.^{18, 19} Specifically, we employed a multicomponent silica nanoparticle that we recently developed.²⁰ The nanoparticle consists of a large mesoporous silica shell and a small iron oxide core. As a case study, we selected a mouse model of malignant brain tumor, which is a hard-to-reach and lethal cancer type. Brain tumors present enormous challenges to drug delivery primarily due to the blood-brain barrier (BBB). However, even though the BBB significantly limits the effective penetration of most hydrophilic drugs into intracranial tumors,^21–25 various studies have shown that the partially compromised nature of BBB permits the deposition of nanoparticles in perivascular regions of brain tumors.^26–31 While typical targeting strategies seek to enhance nanoparticle uptake by cancer cells, we sought to specifically guide the nanoparticles to the vascular and perivascular areas of brain tumors, which are more accessible regions for nanoparticles considering the BBB limitations and the high interstitial fluid pressures of tumors. To target the nanoparticles, we used two different peptides, cRGD and CREKA, that target different vascular and near-perivascular biomarkers of the brain tumor microenvironment. Specifically, cRGD was selected because it targets α_vβ₃ integrin, which is an upregulated receptor on the remodeled vascular walls of gliomas.^{10, 31–35} CREKA targets fibronectin, which is an overexpressed extracellular biomarker in the perivascular space of gliomas.^36–40 It should be noted that these biomarkers are highly suitable for directing nanoparticles to brain tumors as they are highly targetable with minimum off-target occurrences in healthy brain tissues.^{31, 36}

We compared the intratumoral deposition of α_vβ₃ integrin-targeting MSN, fibronectin-targeting MSN and untargeted MSN. The nanoparticles were loaded with the cytotoxic drug Doxorubicin (DOX). While Temozolomide is the first-line chemotherapy used in brain tumors, DOX-loaded nanoparticles have shown excellent anticancer efficacy against brain tumors in preclinical and clinical studies. ^{20, 31, 41} In addition to its cytotoxic potency, DOX exhibits mild fluorescence, which allows direct quantitative analysis in tumors and organs after the drug is extracted from tissues using an established procedure.^{31, 42, 43} Using different doses within a commonly used range of nanoparticles loaded with chemotherapeutic agents (i.e., 2-8 mg of drug per kg of body weight), we identified that each targeted nanoparticle variant reached its upper threshold at a different dose. Using an injection cocktail containing both the α_vβ₃ integrin-targeting MSN and fibronectin-targeting MSN, we show that the intratumoral level of the drug was significantly increased by combining the optimal dose for each nanoparticle variant, which could not be achieved by any of the targeted nanoparticle variant alone. In a subsequent study, we then evaluated the therapeutic efficacy of the strategy. The iron oxide core of the MSN allows drug to be released upon command using a mild low-power radiofrequency (RF) field. As a result of the RF, the iron oxide core vibrates resulting in rapid liberation of drug from the near-perivascular regions, where the nanoparticle deposits, to deeper regions with glioma cells. The combination of higher intratumoral accumulation and the subsequent RF-triggered drug release facilitated improved therapeutic outcomes.

EXPERIMENTAL SECTION

Synthesis of nanoparticles.

The nanoparticles were synthesized following an established method.²⁰ Iron oxide cores were synthesized by the coprecipitation method. First, 0.6757 g of FeCl₃.6H₂O and 0.2478 g of FeCl₂.4H₂O were dissolved in 5 mL of deoxygenated water. Then, 2.5 mL of 0.4 M HCl was added under vigorous stirring. After adding 25 mL of 0.5 M NaOH to the iron solution, mixture was stirred for 15 minutes at 80 °C under argon. The black precipitate was separated by using a powerful magnet followed by several washes with DI water until stable ferrofluid was obtained. To stabilize the nanoparticle suspension, 10 mL of DI water containing 170 mg of citric acid was added and allowed to react at 80 °C for 1.5 hours under argon. Prior to heating, the pH of the reaction mixture was adjusted to 5.2. Finally, the citric acid -coated iron oxide cores were removed by repeated washes and centrifugation. Excess citric acid was removed using Amicon Ultra-15 centrifugal filters.

The final MSN nanoparticles were developed by utilizing a base-catalyzed sol–gel process to generate a large mesoporous silica shell around the iron oxide core. Specifically, 50 mg of the iron oxide core was dispersed in a 25 mL solution of 80% ethanol by ultra-sonication. The iron oxide core suspension was mixed with 5 mL of DI water containing 1 g of cetyltrimethylammonium bromide (CTAB) for 30 minutes. The mixture was then heated to 60 °C for 20 minutes to evaporate ethanol. The Fe₃O₄/CTAB suspension was added to a mixture of 45 mL of water with 0.3 mL of 2M NaOH. After the suspension was heated to 70°C under stirring, 0.5 mL of tetraethylorthosilicate (TEOS) was added under vigorous stirring. At the same time, 54 μL of trihydroxysilylpropylmethylphosphonate was added. After 10 min, 3.3 mg of silane-PEG-NH₂ was added and allowed to react for 24 h at room temperature. CTAB was extracted by refluxing the nanoparticles at 60 °C for 3 h with acidic ethanol at pH ~ 1.4). Any unreacted species were removed from the nanoparticles by washing with ethanol 3 times and repeated centrifugation.

For drug loading, 10 mg of MSN nanoparticles were suspended in 2 mL of PBS at pH of 7.4. A solution of 5 mg DOX.HCl in 1 mL DI water was added to the MSN suspension. The mixture was mildly stirred for 12 h. Free drug was separated by repeatedly washing the nanoparticle suspension with PBS and centrifugation. The drug remaining in the MSNs was then measured. The washing solutions of PBS were analyzed for the remaining drug content by UV-Vis absorption spectroscopy at λ=480 nm.

The RGD or CREKA peptide was conjugated on the surface of the MSN nanoparticle via its distal end of PEG-NH₂ using maleimide chemistry. First, the MSN suspension was mixed with 10 molar excess of sulfo-SMCC (sulfosuccinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate) in PBS for 15 minutes. Next, 5 molar excess (relative to the number of amines on the MSN nanoparticles) of the peptide was added and reacted for 2 h. The nanoparticles were then dialyzed against PBS using a 100,000 Da MWCO dialysis membrane to remove any unreacted peptides. Bio-Rad DC protein assay was used to quantify the number of peptides on the MSN nanoparticles.

Tumor models.

The animal procedures were performed under a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of Case Western Reserve University. The well-being of the animals took priority over completion of the study. Procedures were conducted using anesthesia to minimize pain and distress. For the CNS-1 and GL261 rodent glioma tumor models, 5-8-week-old female athymic nude mice (~25 g) were housed in the Athymic Animal Core Facility according to institutional policies. After the mice were anesthetized by intraperitoneal administration of ketamine and xylazine, 200,000 cells were implanted using into a stereotaxic rodent frame at AP= +0.5 and ML= − 2.0 from bregma at a rate of 1 μL/min in the right striatum at a depth of −3 mm from dura. Mice were then randomized into groups for subsequent studies.

Histological analysis.

Immunohistochemistry was performed to assess the expression of α_vβ₃ integrin and fibronectin and the distribution of drug with respect to glioma cells. Mice were anesthetized with an IP injection of ketamine and xylazine and transcardially perfused with heparinized PBS followed by 4% paraformaldehyde in PBS. Brains were excised, fixed and processed for cryosectioning. Serial 12 μm-thick tissue sections were collected. To visualize the tumor biomarkers of interest, the tissue slices were immunohistochemically stained with anti-α_vβ₃ integrin or anti-fibronectin primary antibody (BD Biosciences, Pharmingen). Green fluorescence was used for imaging the GFP-expressing cancer cells. Direct red fluorescence was used for imaging DOX. Tissue sections were imaged at 10 or 20x on the Zeiss Axio Observer Z1 motorized FL inverted microscope.

Evaluation of nanoparticle deposition in brain tumors.

The intratumoral deposition of RGD-NP, CREKA-NP and the cocktail of RGD-NP and CREKA-NP was assessed in the orthotopic CNS-1 glioma model in mice. After 3 h from tail vein injection of nanoparticles, the mice were anesthetized with an intraperitoneal injection of ketamine and xylazine and transcardially perfused with heparinized PBS. Liver, spleen, kidneys, hearts and brains were then excised, washed, dried, and weighed. The drug was then extracted from the tissues using an established protocol.³¹ Briefly, organs and tumors were homogenized in deionized water (20% wt/vol). A mixture containing 200 mL of the homogenate, 100 mL of 10% Triton X-100, 200 mL of water and 1500 mL of acidified isopropanol (0.75 N HCl) was stored overnight at −20 °C. To separate the extracted drug, the samples were then centrifuged at 15,000 g for 20 min. Total DOX content of the supernatant was analyzed (λ_ex = 480, λ_em = 590) using a fluorescence spectrophotometer (Synergy HT, Biotek). We measured background fluorescence in organs from mice that were injected with a saline solution.

Bioluminescence imaging.

Bioluminescence imaging was performed using an IVIS Spectrum system 10 min after intraperitoneal administration of 200 μl of D-luciferin (10 mg/ml). Animals were imaged every 2 days until the end of the study.

Statistical analysis.

Statistical analysis was performed in Prism version 7 for Mac (GraphPad Software, La Jolla, CA, USA). Data are represented as mean ± s.d. Analysis of differences between two groups was performed using Student’s t-test. Data from three or more groups were analyzed with a one-way analysis of variance (ANOVA) with Sidak’s multiple comparisons test.

RESULTS

Synthesis and characterization of nanoparticles.

The nanoparticles were synthesize following a previously established method.²⁰ Briefly, the iron oxide cores were synthesized using the coprecipitation method. After washing, we coated the iron oxide cores with citric acid. A mesoporous silica shell around the iron oxide core was generated using a base-catalyzed sol–gel process. Initially, the iron oxide cores were mixed in water with water at a pH near 12. After the mixture was heated to 70°C, TEOS was added dropwise and allowed to react for 20 min followed by functionalization with phosphonate. After 2 hours, the nanoparticles were washed twice with ethanol. Surface modification of the nanoparticles was carried out by reacting them with silane-PEG-NH₂ at 70°C for 3 hours. The remaining CTAB was removed in acidic ethanol (pH ~1.5) at 60°C for 3 hours. The nanoparticles were again washed several times in ethanol. Finally, the drug was loaded into the particle using the electrostatic interactions between the drug and mesoporous silica shell. The pH of the nanoparticle suspension was adjusted to 7.4 according to the pKa of DOX. As a result, the positively charged drug molecules exhibited increased electrostatic interaction with the negative charge of the phosphonate-functionalized pores in the silica shell. Free drug molecules were removed by repeated washes with PBS and centrifugation steps.

In Fig. 1a, a TEM image shows the MSN nanoparticles. Size measurements showed that the overall nanoparticle’s size was ~74 nm with the iron oxide core being ~12 nm (Fig. 1b). In previous studies,²⁰ we performed detailed characterization of the nanoparticle including hydrodynamic size and zeta potential measurements, elemental analysis and FT-IR analysis. Upon conjugation of the fibronectin-targeting peptide (CREKA) or the α_vβ₃ integrin-targeting peptide (cRGD) to the amine on the distal end of PEG, each nanoparticle contained ~3,000 peptides as measured by a Bio-Rad DC protein assay. The amount of drug cargo was ~195 pg per 1 mg of nanoparticles (Fig. 1c). Fig. 1d shows that the DOX loading was stable.

Figure 1. Characterization of the nanoparticles.

(a) TEM of the nanoparticles. (b) Size measurements of the MSN nanoparticle and its iron oxide core were obtained from analysis of TEM images (count was 100 particles; data presented as mean ± standard deviation). (c) The MSN nanoparticles were stably loaded with a chemotherapeutic drug (DOX). (d) The stability of drug loading was evaluated in PBS at 37 °C.

Histological evaluation of the targeted biomarkers.

We have previously shown thatfibronectin or α_vβ₃ integrin can be used as targetable biomarkers to effectively home nanoparticles to brain tumors in orthotopic GBM models in mice.^{31, 36} Here, we histologically evaluated the expression of both α_vβ₃ integrin and fibronectin in the same brain tumor tissue using an orthotopic CNS-1 model in mice, which is a very invasive rodent glioma model. Fig. 2 shows representative images indicating abundance and selective expression of α_vβ₃ integrin (left panel) and fibronectin (right panel) on the endothelium and around the near-perivascular regions of brain tumors. Both biomarkers were not present in perivascular regions of healthy brain tissues.

Figure 2.

Histological evaluation of brain tumor tissues shows the expression of α_vβ₃ integrin and fibronectin in orthotopic CNS-1 brain tumors in mice. Fluorescence microscopy images were obtained at 20x magnification. The topology of glioma cells is shown with respect to the expression of α_vβ₃ integrin (left panel) and fibronectin (right panel).

Targeting of nanoparticles to brain tumor.

The intratumoral deposition of the untargeted MSN nanoparticles (UT-NP) and their two targeting variants (CREKA-NP and RGD-NP) was evaluated in mice bearing intracranial CNS-1 tumors as a function of dose (n=5 mice per group). In a previous study,²⁰ it was shown that targeting of perivascular biomarkers, such as fibronectin and α_vβ₃ integrin, enabled MSN particles to rapidly deposit to brain tumors. Specifically, the intratumoral deposition of targeted nanoparticles in brain tumors plateaued to its maximum value of 4.7% the injected dose within 3 h after tail-vein injection. We also show the organ distribution and tumor deposition of the untargeted nanoparticle at 3 and 24 h (Supporting Fig. S1). Thus, we elected to measure the intratumoral drug levels at 3 h after tail-vein injection of the targeted nanoparticles. The dose range for each nanoparticle variant was from 2 to 8 mg/kg DOX. Organs and brain tumors were perfused and excised and drug levels were directly measuring in homogenized tissues using an established method.³¹ We used the drug levels in tissues as a measure of the nanoparticle deposition.

Fig. 3a shows that the deposition of CREKA-NP in brain tumors increased significantly by escalating the dose from 2 to 5 mg/kg. However, the intratumoral accumulation reached a plateau at the higher dose of 8 mg/kg, indicating that the upper threshold of the dose for CREKA-NP is ~5 mg/kg. In a previous study,²⁰ the CREKA-NP injected at 5 mg/kg achieved a 23-fold higher deposition in brain tumors than the standard unmodified DOX. In the case of RGD-NP, there was no benefit to the intratumoral drug levels when we increased the dose from 2 to 5 mg/kg, which suggests that RGD-NP exhibited a lower threshold at 2 mg/kg. Both targeted NP variants exhibited a nearly 2-fold higher intratumoral deposition than the untargeted NP administered at 5 mg/kg. We then tested whether a combination of both targeted nanoparticle variants administered at the upper dose threshold for each nanoparticle can achieve an increased intratumoral deposition. A group of animals was injected with a cocktail containing CREKA-NP at 5 mg/kg and RGD-NP at 2 mg/kg resulting in a total dose of 7 mg/kg DOX. Fig. 3b shows that the combination of the two targeted NP variants achieved an approximately 3-fold increase compared to the upper dose threshold of any formulation. In addition to the total amount of drug in tumors, we also computed the percent of the injected dose that accumulated in the tumor (Fig. 3c). For example, the cocktail of nanoparticles resulted in 4.6 % of its injected dose being deposited in the tumor. While the cocktail clearly outperformed all formulations and doses with respect to the total amount of intratumoral drug, it was injected at a dose of 7 mg/kg. On the other hand, when RGD-NP was injected at a dose of 2 mg/kg, its intratumoral deposition was 5.9% of the dose, even though the total amount of drug in the tumor was half of the cocktail.

Figure 3. Deposition of Dox incorporated in targeted and untargeted MSN nanoparticles in brain tumors.

(a) The MSN formulations were intravenously injected in mice bearing intracranial CNS-1 tumors at day 8 after tumor inoculation. The α_vβ₃ integrin-targeting MSN nanoparticles (RGD-NP) were administered at doses of 2 or 5 mg/kg, whereas the fibronectin-targeting MSN nanoparticles (CREKA-NP) were given at doses of 2, 5 or 8 mg/kg. The dose of the untargeted MSN nanoparticles (UT-NP) was 5 mg/kg. As a measure of the nanoparticle deposition, the drug levels in tissues were quantified 3 h after injection. (b) Another group of mice was injected with a cocktail containing RGD-NP and CREKA-NP injected at a dose of 2 and 5 mg/kg, respectively. (c) Based on the dose of its formulation and its tumor deposition, the % injected dose was calculated. Grouped analysis ANOVA with Sidak’s multiple comparisons test (n=5 mice per group). P values: a *0.0288 and *0.0365, ****<0.0001, b *0.0143 and *0.0187, ns=not significant.

Organ deposition.

The clearance of the MSN nanoparticles from blood circulation was dominated by the reticuloendothelial system (RES), which is the common fate for most types of nanoparticles. The majority of the non-specific deposition of the nanoparticles was found in the liver and spleen (Fig. 4). The amounts of DOX in the kidney, lungs and heart were very low compared to drug found in RES organs. As expected, the clearance of the particles by the liver and spleen was dose-dependent. For example, the deposition of RGD-NP in the liver was 2.53-fold higher for the 5 mg/kg dose than the 2 mg/kg dose. Similar trends were observed for CREKA-NP. Notably, CREKA-NP exhibited lower uptake by the liver compared to equivalent doses of the RGD-NP. On the other hand, the uptake of the nanoparticles by the spleen was very low compared to liver. An exception was the high spleen deposition of the cocktail containing RGD-NP and CREKA-NP, which was about 4-fold greater than all the other conditions.

Figure 4. Distribution of Dox incorporated in targeted and untargeted MSN nanoparticles in the liver and spleen of mice bearing intracranial CNS-1 tumors.

The α_vβ₃ integrin-targeting MSN nanoparticles (RGD-NP) were administered at doses of 2 or 5 mg/kg, whereas the fibronectin-targeting MSN nanoparticles (CREKA-NP) were given at doses of 2, 5 or 8 mg/kg. The dose of the untargeted MSN nanoparticles (UT-NP) was 5 mg/kg. Another group of mice was injected with a cocktail containing RGD-NP and CREKA-NP injected at a dose of 2 and 5 mg/kg, respectively. Organ distribution of all the nanoparticle formulations was assessed 3 h after injection. Grouped analysis ANOVA with Sidak’s multiple comparisons test (n=5 mice per group). P values: **0.00158, ****<0.0001, ns=not significant.

Evaluation of therapeutic efficacy.

So far, we have shown that targeting of biomarkers in the near-perivascular regions of tumors resulted in significantly high drug levels. However, the intratumoral deposition of the nanoparticles in perivascular areas did not allow them to have direct access to cancer cells deep in the interstitial space of brain tumors. To rapidly release the content of the MSN nanoparticles and allow the drug molecules to diffuse and spread deep in the interstitial space of brain tumors, we applied an external low-power radiofrequency (RF) field. The RF system includes electromagnets and audio power amplifiers with a power of ~30 Watts and output of ~5 mT (frequency f=50 kHz). The coil is custom-made with a solenoid’s resistance of ~5 Ohms (N=105 turns, inner diameter=4 cm). In a previous study,²⁰ we performed detailed mechanistic studies that showed that the iron oxide core of the MSN nanoparticle vibrates under the alternating magnetic field. This vibration causes release of the drug molecules from the pores of the MSN nanoparticle, which results in spread of drug throughout the interstitial space of tumors.²⁰ Fig. 5a shows the effect of the RF field (50 kHz) on drug release from the MSN nanoparticles in vitro, indicating rapid release of the majority of the drug cargo within 1 h.

Figure 5. Evaluation of the anticancer effects of the various targeted MSN treatments.

(a) The effect of radiofrequency on triggering drug release from MSN nanoparticles was evaluated in vitro at 50 kHz (~5 mT). After exposure to RF for different amounts of time, fluorescence spectroscopy was used to measure DOX release (n=3; unpaired two-tailed t-test; P values: ***<0.001). (b) Various DOX-loaded MSN nanoparticles were systemically injected into mice bearing intracranial GL261 tumors on day 6, 7 and 10 after tumor inoculation. Treatments included RGD-NP (2 mg/kg), CREKA-NP (5 mg/kg) or a cocktail containing RGD-NP (2 mg/kg) and CREKA-NP (5 mg/kg). When the external RF was used, the head of the mouse was placed in the center of the RF coil 30 min after administration of the MSN nanoparticles. A RF field was applied for 60 min. The GL261 cells stably expressed firefly luciferase, which enabled longitudinal bioluminescence imaging (BLI) to monitor the disease progression. When animals showed a 10% loss of body weight, they were euthanized in a CO₂ chamber, which was the primary endpoint criterion (# indicates the time point when more than 60% of group was not alive). Quantification of the BLI light emission from the whole head is shown (n=5 mice in each group; P values: *<0.05, ****<0.0001)

The therapeutic efficacy of the various DOX-loaded MSN nanoparticles was assessed in an orthotopic rodent GL261 glioma model.⁴⁴ For the targeting studies, we used the CNS-1 mouse model due to its invasive nature, which makes it one of the most hard-to-reach orthotopic brain tumor models. In this study, the GL261 was used, because it is one the standard GBM models in immunocompetent mice to evaluate therapeutic efficacy of targeted nanoparticles.^{20, 31, 36} The GL261 glioma cells stably expressed firefly luciferase allowing us to monitor the response of tumors to the treatments using longitudinal bioluminescence Imaging (BLI) in vivo. (Fig. 5b). The treatments included RGD-NP (2 mg/kg), CREKA-NP (5 mg/kg) or a cocktail containing RGD-NP (2 mg/kg) and CREKA-NP (5 mg/kg). Treatments started on day 6 after tumor inoculation, when the tumor size was ~1 mm. The formulations were intravenously injected three times on day 6, 7 and 10 after tumor inoculation. It is important to emphasize that most GBM models in mice exhibit a very short lifespan. Specifically, mice bearing GL261 brain tumors typically do not survive for more than 10 days. The treatment with RGD-NP or CREKA-NP had negligible therapeutic benefits with their BLI curves being similar to that of the untreated group. This is not surprising, because even though the nanoparticles were deposited in high numbers in perivascular regions of the brain tumor, the drug concentration was below its cytotoxic levels far away from the nanoparticle. Mice treated with the free ‘unmodified’ drug followed by RF showed similar progression of the disease as the untreated mice. After the RF field was applied for 60 min, the animals treated with RGD-NP or CREKA-NP exhibited an improved outcome compared to the same treatments without RF. We then injected a group of mice with the cocktail containing RGD-NP and CREKA-NP at their optimal doses (2 and 5 mg/kg, respectively). Notably, the treatment with the nanoparticle cocktail (+RF) resulted in superior therapeutic outcome than the CREKA-NP (+RF) or RGD-NP treatment (+RF). This is in good agreement with the greater intratumoral deposition of the nanoparticle cocktail compared to RGD-NP or CREKA-NP as shown in Fig. 3. Examples of the drug spread from the MSN nanoparticle (with or without RF) is shown using histological analysis of brain tumor tissues (Supporting Fig. S2).

DISCUSSION

There are many factors that govern the intratumoral deposition of nanoparticles. Many studies have focused on the effect of size and PEG coating of DOX-loaded nanoparticles on the particle’s blood circulation, tumor deposition and anticancer activity.^45–55 These studies concluded that the size of the nanoparticle is a key factor that determines blood residence time, which in turn relates to tumor deposition and tumor retention.⁵¹ In fact, previous studies focusing on liposomes determined that PEGylated unilamellar liposomes with a diameters ranging from 50-150 nm exhibited prolonged circulation time resulting in elevated intratumoral accumulation and antitumor activity.^{48, 51} This size range was identified as a compromise between drug payload (increases with increasing size) and intratumoral accumulation (decreases with increasing size). However, careful analysis of those studies shows variable in vivo performance with significant overlap between the in vivo efficacy of nanoparticles of different sizes. While nanoparticles localize themselves with high specificity in solid tumors due to the so-called enhanced and permeation effect, several features of the tumor microenvironment (TME) govern intratumoral delivery, including blood vessel density and architecture, hemodynamics and interstitial pressures. Considering the highly heterogeneous TME even across adjacent cancer cells, the intratumoral accumulation of nanoparticles strongly depends on the individual tumor being treated,⁵⁶ which is consistent with the broad variation in observed efficacy of nanoparticles in animal studies and clinical experience.

Besides the physicochemical properties (e.g., size, shape, charge), the surface of the nanoparticle can be decorated with a ligand that targets an upregulated receptor in tumors. Similar to the passive intratumoral accumulation, previous studies have shown a strong relationship of the size of targeted nanoparticles and their ability to effectively target overexpressed receptors in tumors.⁵⁶ The purpose of our study was not to further evaluate the role of the physicochemical aspects of nanoparticle design. Thus, for the targeted particle in this study, we selected a commonly used size by synthesizing a mesoporous silica nanoparticle with a diameter of ~80 nm. We derived our targeting scheme based on two established biomarkers of the disease. First, it is recognized that α_vβ₃ integrin is highly overexpressed on the angiogenic endothelium of solid tumors, including brain tumors.^{10, 31–35} Our second target involved fibronectin and fibrinogen, which are not native matrix proteins in healthy brain tissue. On the other hand, the near-perivascular areas of tumors contain a fibrin meshwork due to leakage of plasma fibronectin through the tumor’s leaky endothelium.^36–40 Our histological analysis confirmed the overexpression of α_vβ₃ integrin and fibronectin in the two mouse models of GBM we used in this study.

In previous studies,^57–60 we employed multi-ligand targeting strategies to show that the dynamic tumor microenvironment displays a continuously changing expression of targetable receptors over space and time. Those studies showed that each ligand on the nanoparticle was able to target a different region within the same tumor. As a result, a dual-ligand nanoparticle captured the spatiotemporal changes of the receptors throughout the TME allowing accurate detection of breast cancer (BC) metastasis disease using PET imaging.⁵⁷ It should be emphasize that our multi-targeting strategy was not designed to target simultaneously two receptors on the same nanoparticle.^{61, 62} In a follow-up study, we expanded the targeting scheme to 4 ligands, which facilitated detection of dormant BC metastatic disease or early aggressive metastasis using PET.⁵⁸ Those earlier studies focused on employing multi-ligand nanoparticles to ensure that no region with cancer cells was missed for diagnostic purposes.^57–59 While multi-ligand nanoparticles exhibited accurate diagnosis, it is challenging to control the ‘dose’ of each ligand, because it requires carefully tuning of each ligand density on the same nanoparticle.¹⁷ In the current study, our first objective was to initially identify the highest dose for each nanoparticle variant beyond which there is no additional benefit in terms of intratumoral accumulation. Even though our dose escalation only used 3 different levels (2.5, 5 and 8 mg/kg), the two variants exhibited a very different upper limit dose. Our second objective was to evaluate whether administration of both targeted nanoparticles together at their optimal dose could overcome the highest intratumoral accumulation of the individual targeted nanoparticle variants when injected alone. In fact, the animal studies showed the combination of the two targeted NP variants resulted in a 3-fold increase of the intratumoral drug accumulation compared to any formulation administered at similar doses. These data are consistent with our previous findings, where multi-ligand nanoparticles achieved significantly greater intratumoral deposition than their single-ligand counterparts.^57–60 Considering that each ligand is overexpressed at different levels in different regions of the same tumor, it is not surprising that the injection of two nanoparticles that target two different biomarkers resulted in higher accumulation in regions that expressed both biomarkers or only one of them.

After the targeting studies, we sought to evaluate the therapeutic efficacy of the combination of two targeted nanoparticles administered at their optimal doses. Considering that nanoparticles deposit in the perivascular regions of brain tumors,³⁰ we used a low-power RF field to trigger drug release from the nanoparticles and allow the drug molecules to diffuse into the tumor tissue. In a previous study,²⁰ we showed that the RF-triggered release mechanism is governed by the Brownian relaxation of the iron oxide core, which forces the vibration of the entire nanoparticle giving sufficient energy to the drug molecules to overcome the electrostatic interactions that keep them trapped in the mesoporous silica component of the nanoparticle. Due to the very long wavelengths of the selected RF field, release can be triggered at any tissue depth.

We realize that the variable cytotoxic effect of nanoparticles can be attributed to chemoresistance to the administered drug. However, the variable responsiveness to chemotherapy has also been linked to the amounts of drug reaching the tumor.⁶³ In conclusion, we show that a higher dose of a targeted nanoparticle does not guarantee greater drug accumulation in the tumor but contributes to elevated off-target deposition and potentially more serious toxicities. By strategically combining two different targeting schemes and selecting the appropriate dose for each, targeted nanoparticles can facilitate remarkable increase in intratumoral drug delivery that is not achievable by a sole targeting nanoparticle alone.

Supplementary Material

Supporting Information

Fig. S1 shows the deposition of untargeted nanoparticles in brain tumors, liver and spleen.

Fig. S2 shows histological assessment of the RF-triggered release of DOX from nanoparticles in vivo.