Combined Gadolinium and Boron Neutron Capture Therapies for Eradication of Head-and-Neck Tumor Using Gd¹⁰B₆ Nanoparticles under MRI/CT Image Guidance

Abstract

Eradication of head-and-neck (H&N) tumors is very difficult and challenging because of the characteristic feature of frequent recurrence and the difficulty in killing cancer stem cells. Neutron capture therapy (NCT) is emerging as a noninvasive potential modality for treatments of various types of tumors. Herein, we report that 98.5% ¹⁰B-enriched anti-EGFR-Gd¹⁰B₆ nanoparticles can not only deliver large doses of 158 μg ¹⁰B/g tumor tissues as well as 56.8 μg ¹⁵⁷Gd/g tumor tissues with a very high tumor-to-blood (T/B) ¹⁰B ratio of 4.18, but also exert very effective CT/MRI image-guided combined GdBNCT effects on killing cancer stem cells and eradication of recurrent head-and-neck (H&N) tumors. This leads to a long average half-lifespan of 81 days for H&N tumor-bearing mice, which is a record-making result, and surpasses the best result reported in the literature using combined radiotherapy and T cell-mediated immunotherapy (70 d).

Introduction

Cancer is one of the deadliest diseases leading to an estimated 9.6 million deaths worldwide in 2018, of which 5% were due to recurrent head-and-neck (H&N) tumors.^1,2 Despite that many medical modalities have been developed to treat H&N cancers, recurrence of H&N tumors always occur frequently. Inhibiting recurrence of H&N tumors remains as one of the toughest therapeutic challenges.^3,4 Boron neutron capture therapy (BNCT) has recently emerged as a noninvasive, alternative modality for treating various kinds of tumors. The concept of neutron capture therapy is to utilize thermal neutron capture elements, such as ¹⁰B or ¹⁵⁷Gd, to capture a low energy (∼0.025 eV) thermal neutron beam and initiate a nuclear fission reaction to generate high-energy α-/⁷Li-particles from ¹¹B or γ-ray/Auger electrons from ¹⁵⁸Gd* to kill cancer cells.^5,6 Traveling of high-energy α-/⁷Li-particles leads to damage to nearby ∼1–2 cells within the ∼9–14 μm range. Physical damage of cancer cell membranes and cellular components by high-energy α-/⁷Li-particles do not rely on the presence of molecular oxygen, and thus, BNCT can be used to kill cancer cells in hypoxic tumor domains. In the case of GdNCT, the γ-ray emitted from ¹⁵⁸Gd* can travel several centimeters to cause damage to cancer cells in much larger volumes of tumor tissues, which can compensate the limitations of a very short killing zone of BNCT. Ideally, both BNCT and GdNCT are very good noninvasive modalities for treating various types of cancers.^3,7 The development of BNCT and GdNCT, however, is currently hindered by the lack of proper reagents able to deliver sufficiently large amounts of ¹⁰B and/or ¹⁵⁷Gd atoms to tumor sites to fulfill theoretical threshold values of >20–30 μg ¹⁰B/g tumor tissue and a tumor-to-blood (T/B) boron ratio of >3.0 to exert effective BNCT, or 50–200 μg ¹⁵⁷Gd/g tumor tissue⁸ for effective GdNCT therapy on killing cancer cells. Currently, the most effective and commonly used molecular BNCT drugs are 4-borono-l-phenylalanine (BPA) and mercapto-undecahydrododecaborate (BSH).^8,9 Molecular drug-based BNCT and GdNCT were also used to treat recurrent H&N tumors.¹⁰⁻¹³ However, in most of cases, modest medical efficacies were observed with frequent recurrence of H&N tumors within 2 years after BNCT treatment, along with some adverse effects, such as mucositis, oral pain, fatigue, etc. In the literature, it was attempted to conduct combined GdBNCT on killing cancer cells using poly(lactic-co-glycolic acid) (PLGA) polymeric nanoparticles-delivered molecular ¹⁵⁷Gd-complex and B-containing curcumin.¹⁴ Experiments were performed at the cellular level, with no data presented at the in vivo solid tumor level. Modest GdBNCT was observed, which was most probably due to the very low Gd content (one ¹⁵⁷Gd atom out of 158 atoms) and B content (one ¹⁰B atom out of 53 atoms) in these molecular complexes, without considering the dilution effect from the PLGA polymeric nanoparticles carriers. In other studies, a Gd-orthocarborane complex, called AT101, was used to conduct MRI-guided BNCT for the treatment of transplanted mesothelioma tumors using low density lipoproteins (LDL) or β-cyclodextrin (β-CD) as nanocarriers. At a time point of 3 h post i.v. injection of the AT101-LDL complex nanoparticles, a high quantity of 36 μg ¹⁰B/g tumor tissue was observed. After thermal neutron irradiation, a significant tumor size reduction was observed.¹⁵⁻¹⁷ The amounts of ¹⁵⁷Gd and ¹⁰B atoms in the above AT101 adduct, however, are quite low, i.e., 1 ¹⁵⁷Gd atom and 10 ¹⁰B atoms out of total 154 atoms in the adduct. If the atomic percentages of both Gd and B atoms can be boosted further, then excellent BNCT therapeutic efficacy can be expected. Overall, most small molecule-based reagents show modest BNCT and GdNCT efficacies.¹⁸⁻²⁶Table S1 provides an overview of various Gd-containing nanocarriers used for GdNCT treatments for different types of tumors.^{18−24,27,28} The main challenges and limitations for BNCT and GdNCT are (i) due to the very short travel distance of ∼9–14 μm ± 3 μm, α and ⁷Li particles generated from BNCT can only kill nearby cancer cells, but not those in remote tumor tissues. Uneven distribution of BNCT reagents in tumor tissues and incomplete eradication of cancer cells, especially cancer stem cells, in the remote tumor tissues very often lead to recurrence of H&N tumors. (ii) Due to the lack of tumor-targeting ability and low ¹⁰B and ¹⁵⁷Gd contents, it is very difficult for molecular BNCT and GdNCT prodrugs to fulfill the above-mentioned threshold values and exert effective NCT on killing cancer cells. (iii) Boron atom does not have magnetic property, and it is difficult to use noninvasive bioimaging methods (such as MRI or CT) to monitor the in vivo biodistriubution/concentrations of BNCT reagents and identify the location and the size (and thus the growth) of a tumor. Although radioactive ¹⁸F can be chemically introduced into BPA to form radioactive ¹⁸F-BPA which can be monitored using positron emission tomography, the very short radiation half-life of 109.8 min makes it very difficult for the immediate prior-use synthesis, delivery, storage, and biomedical applications. In addition, the high-energy positron emission from ¹⁸F can also cause damage to normal human tissues/organs.

To conquer the above-mentioned BNCT- and GdNCT-related problems and eradication of recurrent H&N tumor all at once, we present Gd¹⁰B₆ NPs as dual modal magnetic resonance imaging/computor tomography (MRI/CT) reagents as well as prodrugs to deliver sufficiently large amounts of boron and gadolinium elements to a tumor site and to exert effective combined GdBNCT for destroying H&N tumors. On top of BNCT, an additional GdNCT can provide high-energy γ-ray to kill those cancer cells in distant tumor tissues. The magnetic Gd element can also serve as a stable, noninvasive bimodal MRI and CT contrast agent to help determine the biodistribution (and possibly the concentrations as well) of both Gd and B elements as well as identify the location, the size, and the growth of solid tumors and avoid the use of troublesome short-lived radioactive ¹⁸F labeling. Gd¹⁰B₆ NPs can carry large amounts of both Gd and B atoms to fulfill simultaneously the threshold conditions for effective GdBNCT. Our experimental results show that combined GdBNCT using anti-EGFR-Gd¹⁰B₆ NPs could destroy H&N solid tumor, including the cancer stem cells, more effectively as compared to pure BNCT (using anti-EGFR-Ce¹⁰B₆ NPs or BPA-F), pure GdNCT (using Gd₂O₃ NPs), and DOX-based chemotherapy. The mice group treated with combined GdBNCT exhibited a very long average half-life span of 81 d, far exceeding that of mice treated with pure BNCT using anti-EGFR-Ce¹⁰B₆ NPs+NR (60 d), anti-EGFR Gd₂O₃ NPs + NR (52 d), BPA-F 1 h+NR (38 d), BPA-F 24 h+NR (34 d), DOX-treated group (28 d), control+NR (24 d), and control (21 d). We also demonstrate that combined GdBNCT could kill cancer stem cells effectively and result in nearly complete eradication of recurrent H&N tumor, which is far better than pure BNCT and DOX-based chemotherapy.

Result and Discussion

Preparation and Characterization of Gd¹⁰B₆ and Ce¹⁰B₆ NPs

Synthesis of natural abundance metal hexaborides using high temperature furnance methods has been well reported in the literature. These literature methods are not suitable for the synthesis of ¹⁰B-enriched metal hexaborides due the requirement of large quantities of expensive ¹⁰B-enriched precursors. In this study, we developed a unique modified microwave arcing method²⁹⁻³¹ to prepare ¹⁰B-enriched Gd¹⁰B₆ NPs using GdI₃, ¹⁰B-enriched boric acid (H₃¹⁰BO₃ isotropic purity ≥98.5% ¹⁰B atom%, Santa Cruz Biotechnology, Dallas, Texas, U.S.A) and magnesium (Mg) metal pieces as precursors. Briefly, in a focused microwave oven, the Mg metal pieces absorb microwave and generate violent microwave arcing, within which the GdI₃ and H₃¹⁰BO₃ were broken down into atomic plasma and fragments, such as Gd, O, H, ¹⁰B, etc. Upon cooling, all atomic fragments will self-assemble into corresponding thermodynamically stable products. Gd¹⁰B₆ nanoparticles are one of the final thermodynamically stable products (see Figure 1a). The Gd¹⁰B₆ NPs were purified by using HCl_(aq) aqueous solution to dissolve other basic metal oxides (such as Gd₂O₃ and MgO). The as-prepared Gd¹⁰B₆ NPs were characterized by various spectroscopic techniques, including transmision electron microscopy (TEM), scanning electron microscopy (SEM), X-ray diffraction (XRD) pattern, and X-ray photoelectron spectrometry (XPS) and UV–vis-NIR absorption spectrum (see Figure S1 in SI for detail characterization). It was reported that epidermal growth factor receptors (EGFR) are overexpressed in more than 90% of H&N tumor patients.³² To introduce cancer cells targeting ability, the as-prepared Gd¹⁰B₆ NPs were coated with a layer of F127-COOH polymer, via electrostatic interactions of the caboxylate anion of F127 and Gd³⁺ cations, followed by further surface-functionalized with anti-EGFR antibody via 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide (EDC) coupling of the amino group in anti-EGFR moieties and the carboxylic acid group in the F127-COOH polymer chains.³³ The final nanoparticles were designated as anti-EGFR-Gd¹⁰B₆ NPs (Figure 1b). The surface functional moieties on anti-EGFR-Gd¹⁰B₆ NPs were characterized by the ¹³C NMR spectrum (Figure S2) and the FT-IR spectrum (Figure S3). The magnetic susceptibility χ(T) measurements showed that anti-EGFR-Gd¹⁰B₆ NPs have a saturated magnetization value of 0.0062 emu/g at 2 K and an external magnetic field of 7T in the absence of a hysteresis loop (Figure S4). The results indicated that anti-EGFR-Gd¹⁰B₆ NPs are superparamagnetic,³⁴ that is, when the external magnetic field was removed, the net magnetic moment of Gd¹⁰B₆ nanoparticles goes back to zero; and the Gd¹⁰B₆ nanoparticles will not stack together to form large aggregates. Anti-EGFR-Ce¹⁰B₆ NPs were prepared and characterized in a similar way to the anti-EGFR-Gd¹⁰B₆ NPs (see Figure S5 and S6 in SI). Gd₂O₃ NPs was purchased from a commnercial source (NRE-3020, Nano Research Elements, < 100 nm) and its detailed characterization are shown in Figure S7 and S8, respectively. The overall experimental design of this study was schematically presented in Figure 1. Due to the magnetic and heavy element properties, Gd¹⁰B₆ NPs can serve as both MRI and CT contrast reagents (vide infra) to help monitoring the biodistribution and pharmacokinetics of both Gd and B elements in a biological system. As mentioned above, the high-energy α and ⁷Li particles generated from the BNCT effect can only travel within ∼9–14 μm ± 3 μm only enough to kill nearby ∼1–2 cancer cells, but not those cancer cells in a remote area, especially in the hypoxic tumor subdomains. While the high-energy γ-ray generated from the decay of the ¹⁵⁸Gd* upon neutron beam irradiation can travel a much longer distance to cause damage to those cancer cells in remote regions, that is, the GdNCT effect has a much larger killing zone than that for the BNCT effect and can compensate for the insufficient ability of BNCT for killing cancer cells in remote tumor tissues (see Figure 1d). One may wonder whether the high-energy γ-ray emitted from the ¹⁵⁸Gd* upon neutron beam irradiation could also cause damage to normal cells/tissues in remote regions. Indeed, such kinds of damage to normal cells/tissues seem to be unavoidable but still acceptable due to the following reasons: (a) the γ-ray emission from the ¹⁵⁸Gd* elements located within tumor tissues radiates outward in three-dimensional space. The farther away the radiation ¹⁵⁸Gd* elements (within tumor tissue), the weaker the radiation intensity will be. In other words, the damages to normal cells located far away from the γ-ray radiation souces are far less than damages to tumor tissues at the close proximity to the γ-ray radiation sources. (b) As compared to conventional radiotherapy where the focused X-ray beam causes damages to normal cells/tissues along the pathways coming in and out of tumor tissues, the damages of normal cells/tissues by γ-ray in GdNCT is only one way out of the tumor tissues with gradually decreasing γ-ray intensities along their ways out of the tumor tissues. Qualitatively speaking, the GdNCT based γ-ray damages to normal cells/tissues are expected to be far less than those caused by focused X-ray beams in conventional radiotherapy.

Figure 1.

Schematic concept and experimental designs of GdBNCT in this study. (a) Preparation of Gd¹⁰B₆ NPs by a microwave arcing method. (b) Introducing tumor-targeting ability of Gd¹⁰B₆ NPs by surface-grafting anti-EGFR moieties. (c) Thermal neutron irradiation of mice bearing HTB-43 H&N tumors, as well as in vivo CT/MRI images. (d) Effective killing zones of BCNT (blue spheres, 9–14 um) and GdNCT (red sphere, a few centimeters).

In Vitro Dark Cytotoxicity of Anti-EGFR-Gd¹⁰B₆ NPs

To determine the biocompatibilities and cytotoxicities of both anti-EGFR-Gd¹⁰B₆ NPs and anti-EGFR-Ce¹⁰B₆ NPs, in vitro stardard MTT assay experiments were performed using the American Food and Drug Administration (FDA)-aproved BPA-F complex (the most effective molecular BNCT drug) as a control for comparison. As shown in Figure 2a, the cell viabilities of HTB-43 cancer cells in the presence of BPA-F (F= fructose), anti-EGFR-Ce¹⁰B₆, anti-EGFR-Gd₂O₃, and anti-EGFR-Gd¹⁰B₆ NPs (all have the same dose of 50 μg/mL) in the dark condition are 87.6 ± 1.7%, 86.2 ± 4.3%, 86.5 ± 1.7%, and 88.5 ± 2.6% (n = 4), respectively, indicating that the dark cytotoxicities of anti-EGFR-Gd₂O₃, anti-EGFR-Ce¹⁰B₆ and anti-EGFR-Gd¹⁰B₆ NPs are compatible to that for the FDA-aproved BPA-F. Low dark cytotoxicities of both anti-EGFR-Ce¹⁰B₆ and anti-EGFR-Gd¹⁰B₆ NPs were similarly observed toward a healthy human umbilical vein endothelial cells (HUVECs) (Figure S12). In sharp contrast to low dark cytotoxicities, upon exposure to neutron irradiation, the number of cell deaths of HTB-43 cancer cells for different treatment groups increases proportionally in a dose-dependent manner. At a dose of 50 μg/mL and a thermal neutron irradiation (neutron flux = 1 × 10⁹ n/cm²sec, 60 min),^35,36 the net neutron irradiation-induced cell deaths are in the following order: 5.7 ± 0.1% (BPA-F) < 12.8 ± 0.6% (anti-EGFR-Gd₂O₃ NPs) < 17.8 ± 0.4% (anti-EGFR-Ce¹⁰B₆ NPs) < 30.9 ± 1.5% (anti-EGFR-Gd¹⁰B₆ NPs) (n = 4, Figure 2b). The much higher neutron irradiation-induced cell deaths of nanoparticles-based NCT drugs as compared to molecular BNCT drug (i.e., BPA-F) are attributed to two factors: namely, (a) the much higher Gd or B atom contents in the nanoparticle-based NCT drugs and (b) the cancer cells targeting abilities of anti-EGFR-M¹⁰B₆ (M= Ce, or Gd) and anti-EGFR-Gd₂O₃ NPs. Molecular drug, BPA-F, does not have cancer cell-targeting ability. The anti-EGFR moieties on the Gd¹⁰B₆ NPs surface dramatically increase their cancer cells’ uptake by 3.2-fold as compared to the uptake of bare Gd¹⁰B₆ NPs by HTB-43 cancer cells (Figure 2b), leading to largely enhanced neutron irradiation-induced cell deaths. Notice that the percentage of cell uptake (Figure 2b) for anti-EGFR-Gd₂O₃ is slightly lower than those for the other two nanoparticles. The molecular weights of Gd₂O₃, Ce¹⁰B₆, and Gd¹⁰B₆ are 362, 200, and 217 g/mol, respectively. Assuming the particle sizes for all three types of nanoparticles are about the same, the number of nanoparticles for Gd₂O₃ will be the least among all under the condition of the same weights being used in the cellular uptake experiments (Figure 2b). The lower number of nanoparticles for Gd₂O₃ under the condition of the same weight could be the reason responsible for the slightly lower cellular uptake shown in Figure 2b. Another important feature in Figure 2a is that combined GdBNCT effects (30.9 ± 1.5% cell deaths from the anti-EGFR-Gd¹⁰B₆ NPs group) could result in much higher neutron irradiation induced cell deaths as compared to either the BNCT effect (17.8 ± 0.4% cell deaths from the anti-EGFR-Ce¹⁰B₆ NPs) or the GdNCT effect (12.8 ± 0.6% cell death from the anti-EGFR-Gd₂O₃ NPs). The data in Figure 2a also showed that α-/⁷Li-particles-based BNCT is more effective in killing immediately nearby cancer cells (17.8 ± 0.4% cell deaths) than the γ-ray-based GdNCT (12.8% cell deaths).

Figure 2.

In vitro cytotoxicities, cell uptake, and cell death pathways of HTB-43 H&N cancer cells. MTT assay for various groups in (a) the dark and (b) under neutron irradiation. (c) Apoptosis and necrosis rates of different treatments. (d) Flow cytometry cell uptake experiments for cancer cells-targeting ability of anti-EGFR-Gd¹⁰B₆ NPs and anti-EGFR-Ce¹⁰B₆ NPs. (e) Fluorescence images of Calcein-AM and PI double staining to visualize HTB-43 cells treated with different groups upon neutron irradiation (scale bars, 50 μm). The error bars represent the standard deviation for four independent measurements. NR= neutron irradiation. The dose of 50 mg/kg of nanoparticles is equivalent to the amounts of (36.15 mg of Gd/kg and 13.85 mg of ¹⁰B/kg) for Gd¹⁰B₆, (15 mg of ¹⁰B/kg) for Ce¹⁰B₆, and (43.35 mg of Gd/kg) for Gd₂O₃ nanoparticles, respectively.

The much more efficient cell killing effects from the nanoparticle-NCT drug groups, as compared to the molecular drug BPA-F group, can also be seen from in vitro Calcein-AM and PI staining experiments. As shown in the Figure 2c, only green fluorescence spots (i.e., alive cells) were observed in the dark. However, under neutron irradiation, large amounts of cell deaths were observed for the groups of HTB-43 H&N cancer cells pretreated with nanoparticle-NCT prodrugs (including anti-EGFR-Ce¹⁰B₆ NPs, anti-EGFR-Gd¹⁰B₆ NPs, and anti-EGFR-Gd₂O₃ NPs), where the amounts of dead cells (i.e., the red fluorescence PI stains) for nanoparticle-NCT prodrugs are clearly more than that for the molecular drug BPA-F treated group.

To gain deeper insights regarding the mechanisms of NCT-induced apoptosis and necrosis cell deaths, after neutron irradiation, the cancer cells were stained with FITC-labeled Annexin V and Propidium Iodide (PI), followed by flow cytometry analysis. The Annexin V staining provides a very sensitive method for detecting cell apoptosis, while PI is used to detect necrotic cells, characterized by the loss of the plasma and nuclear membranes and binding of PI to DNA of dead cells. As shown in the Figure 2e, the major BNCT-induced cell death pathway is the necrosis process, of which the percentages of necrosis induced cell deaths are in the following order: BPA-F+NR groups (25.5%) < anti-EGFR-Gd₂O₃ NPs+NR group (33.2%) < anti-EGFR-Ce¹⁰B₆ NPs+NR (52.8%) < anti-EGFR-Gd¹⁰B₆ NPs+NR (61.1%)(Figure 2c). It can be envisioned that the high-energy α and ⁷Li particles created from BNCT can cause breakage of cell membranes and lead to the necrosis of cancer cells. On top of the BNCT effect on necrosis cell deaths, the additional γ-ray from GdNCT can further damage cellular components, retard the repairment of these partially injured cancer cells by BNCT, and thus increase the amount of necrosis-induced cell deaths.

In Vitro and In Vivo Dual-Modal Computed Tomography (CT) and Magnetic Resonance (MR) Imaging Measurements

Gd₁₀B₆ NPs possess magnetic moments and are expected to act as a promising contrast reagent for MRI and CT bioimaging, similar to Gd₂O₃ NPs.³⁷ The dual imaging capability of Gd¹⁰B₆ NPs was demonstrated in both in vitro and in vivo T₁ MRI and CT imaging experiments. In vitro imaging experiments were performed by taking the CT and MRI images from Eppendorf tubes containing different concentrations of Gd¹⁰B₆ NPs (0.1 to 1 mg/mL) using a H₂O solution as a control for comparison. As shown in Figure 3a, the CT Houns-field unit (HU) values increase gradually upon increasing the Gd¹⁰B₆ NPs concentrations. Further, T₁ weighted MRI images were taken in a similar way. A concentration-dependent increase in the T₁-MR contrast intensities was observed (Figure 3b). We then further took in vivo CT and MRI images from mice that were intravenously injected with anti-EGFR-Gd¹⁰B₆ NPs from the mice tail vein. The CT/MRI images were taken at time points of pre- and postinjection (24 h) of anti-EGFR-Gd¹⁰B₆ NPs (dose = 50 mg/kg). As illustrated in Figure 3c,d, the tumor sites could be clearly visualized (see red circles for both CT and MR images) from mice iv-injected with anti-EGFR-Gd¹⁰B₆ NPs. These results unambiguously indicate that anti-EGFR-Gd¹⁰B₆ NPs hold great potential as a dual modal in vivo MRI/CT contrast agent for diagnosis of tumors.

Figure 3.

In vitro and in vivo CT and MRI images of mice using anti-EGFR-Gd¹⁰B₆NPs. (a) CT images and HU values of anti-EGFR-Gd¹⁰B₆ NPs. (b) MR images and MRI contrast intensity. (c) CT and (d) T₁-weighted MR images of tumors under pre- and post i.v. injection of anti-EGFR-Gd¹⁰B₆ NPs (dose= 50 mg/kg) in female BALB/c mice. Representative images highlighting the tumor (T) site and the main organs (K, L, S and H, denoting kidney, liver, spleen, and heart).

In Vivo GdBNCT Eradication of H&N Tumor

To investigate the biodistribution of ¹⁰B and Gd elements in major organs, anti-EGFR-Gd¹⁰B₆ NPs were injected intravenously into HTB-43 tumor-bearing mice, followed by collection of major organs at time points of 12, 24, and 48 h post i.v. injection, followed by inductively coupled mass spectroscopy (ICP-MS) analysis. As shown in Figure 4a, a large fraction (∼42%) of the iv injected anti-EGFR-Gd¹⁰B₆ NPs accumulated at the H&N tumor sites, thanks to the tumor-targeting ability of the anti-EGFR moieties. Upon injection of nanoparticles into the bloodstream of a mouse, large majority of nanoparticles were removed via different biological clearance pathways, such as hepatic clearance, splenic clearance, renal clearance, etc., dependent on the sizes of nanoparticles.^38,39 In the literature, it was reported that in the absence of surface-chelating of tumor-specific targeting biomarkers on the surface of nanoparticles, nanoparticles lack active tumor-targeting ability. Passive accumulation of nanoparticles at a tumor site via the enhanced permeability retention (EPR) effect leads to a very low fraction of ∼0.7% of nanoparticles accumulated at a tumor site.^40,41 Upon surface chelating of tumor-specific biomarkers to introduce active tumor-targeting ability, the fraction of nanoparticles accumulated at a tumor site can be drastically boosted up to ∼10–20%.⁴²⁻⁴⁶ In the current study, ∼42% of nanoparticles were observed to accumulate at the tumor site, which is significantly higher than those observed in other tumor systems and therefore is considered as “large fraction”. ICP-MS analysis showed that the amounts of boron and gadolinium elements at the tumor site were 158 ± 17 μg ¹⁰B/g tumor tissue and 364.1 ± 115 μg Gd/g (or 56.8 ± 18 μg ¹⁵⁷Gd/g) (n= 3), respectively, at 24 h after iv injection. Note that Figure 4a shows a low ¹⁰B and Gd concentration in the liver, although the liver is one of the major organs for clearance of nanoparticles. In the literature, it was known that nanoparticles with different sizes have different major clearance pathways. Nanoparticles with sizes >200 nm were removed mostly through the splenic clearance, and nanoparticles with sizes within 100–200 nm were removed mostly through the hepatic clearance pathway, whereas nanoparticles with sizes smaller than 10 nm were cleared through a renal clearance channel.⁴⁷⁻⁴⁹ In the current study, the sizes for all three types of nanoparticles are in the range of ∼50–60 nm, of which the hepatic clearance pathway is not the major one for these nanoparticles, which is probably the reason why the accumulation of ¹⁰B and Gd elements in liver is not high compared to those in other major organs as well as the tumor site. The tumor-to-blood (T/B) boron ratio is 4.18 at 24 h post injection. Similar effective accumulation of nanoparticles was observed in the case of anti-EGFR-Ce¹⁰B₆ NPs. 151 μg ¹⁰B/g tumor tissue and 337 μg Ce/g tumor tissue were detected in the tumor tissue (see Figure S13) at the time point of 24 h post injection. Hence, we chose the time point of 24 h post injection to conduct thermal neutron irradiation for all groups in this study, except BPA-F where both 1 and 24 h post injection time points were chosen, since it was reported that the time point for optimal BPA-F tumor accumulation is 1 h post i.v. injection of BPA-F.⁵⁰ 48 Mice were randomly divided into eight groups. The mice were iv injected with anti-EGFR-Gd¹⁰B₆ NPs, anti-EGFR-Ce¹⁰B₆ NPs, anti-EGFR-Gd₂O₃ NPs, and BPA-F (all had the same dose of 50 mg/kg), respectively. After 24 h postinjection, corresponding mice groups were exposed to thermal neutron irradiation (neutron flux = 1 × 10⁹ n/cm²sec) for 60 min. The tumor volumes were continuously monitored for 21 days. The tumor volume plots shown in Figure 4b illustrated that the (anti-EGFR-Gd¹⁰B₆ NPs+NR)-treated mice had the smallest average tumor size (1.96-fold relative to the initial tumor volume at day 0) among all and was far smaller than those for anti-EGFR-Ce¹⁰B₆ NPs+NR (13.5-fold), anti-EGFR-Gd₂O₃ NPs+NR (16.8-fold), BPA-F 1 h+NR (35.1-fold), BPA-F 24 h+NR (40.8-fold), DOX (47.6-fold), control+NR (50.9-fold), and the untreated blank control group (57.5-fold). The body weights of mice for various treatment groups did not show significant changes within 21 days during the therapeutic treatments (Figure S14). The (anti-EGFR-Gd¹⁰B₆ NPs+NR)-treated mice group demonstrated a very long average half lifespan of 81 days, which is far longer than 60 days for the (anti-EGFR-Ce¹⁰B₆ NPs + NR)-treated group, 52 days for anti-EGFR-Gd₂O₃ NPs+NR, 38 days for the (BPA-F 1 h+NR) group, 34 days for the (BPA-F 24 h+NR) group, 28 days for the DOX-treated group, and 21 days for the blank control group. Optical images shown in Figure 4d displayed much better shrinkage of tumor volumes (indicated by red circles) during the treatment with (anti-EGFR-Gd¹⁰B₆ NPs+NR) in comparison with other treatment groups. The above tumor growth curves and average half lifespan data unambiguously demonstrated that (a) combined GdBNCT is superior to either pure BNCT or pure GdNCT effect alone on destroying H&N tumors and prolonging average lifespan; and (b) nanoparticle-based BNCT (from anti-EGFR-Ce¹⁰B₆ NPs+NR) is much better than the molecular drug-based BNCT (from BPA-F+NR). (c) The average half lifespan of 81 d for mice bearing H&N tumors and treated with the combined GdBNCT therapy is also far longer than the best results reported in the literature for H&N tumors-bearing mice treated with combine radiotherapy and T cell-mediated immunotherapy (70 d),⁵¹ as well as 69 d for immunotherapy with TLR agonists and checkpoint inhibitors³⁶ (see a summary in Table S2).^{35−37,47−50}

Figure 4.

In vivo GdBNCT therapy for treating HTB-43 H&N tumors. 48 Mice were divided into eight groups, including a blank control (n = 6), control+NR (n = 6), doxorubicin (DOX, an anticancer chemotherapy drug)-treated group (n = 6), BPA-F 24 h+NR (n = 6), BPA-F 1 h+NR (n = 6), anti-EGFR-Ce¹⁰B₆ NPs+NR (n = 6), and anti-EGFR-Gd¹⁰B₆ NPs+NR (n = 6). (a) Biodistribution of both gadolinium and boron elements in mice injected with anti-EGFR-Gd¹⁰B₆ NPs as a function of time. (b) Tumor growth curves for different treatment groups as a function of treatment days after neutron irradiation. (c) Survival rates of various treated mice groups. (d) Photographs of HTB-43 tumor-bearing mice from eight groups during the 21 days period after treatments. The doses of BPA-F, anti-EGFR-Gd₂O₃ NPs, anti-EGFR-Ce¹⁰B₆ NPs, and anti-EGFR-Gd¹⁰B₆ NPs, are the same, i.e., 50 mg/kg, which is equivalent to the amounts of (36.15 mg of Gd/kg and 13.85 mg of ¹⁰B/kg) for Gd¹⁰B₆, (15 mg of ¹⁰B/kg) for Ce¹⁰B₆, and (43.35 mg of Gd/kg) for Gd₂O₃ nanoparticles, respectively.

In Vivo Dark Cytotoxicity of Anti-EGFR-Gd¹⁰B₆ NPs

To have clinical applications for the treatment of hamun patients, the cytotoxicity of a nanomedicine is one of the major concerns. In the above in vitro cellular cytotoxicity measurements, both anti-EGFR-Ce¹⁰B₆ and anti-EGFR-Gd¹⁰B₆ NPs show compatible (dark) cytotoxicities as that for the FDA approved molecular BNCT drug, i.e., BPA-F (see Figure 2a, above). We further proceed to measure the in vivo cytotoxicities of these two NPs by collecting the serum samples of mice bearing H&N tumors and iv injected with these two NPs. The liver and kidney functions were determined using the standard inflammatory biochemical parameters, where (AST, ALT, ALP, and ALB) biomarkers are used to evaluate the liver functions, while (BUN and CRE) are used as biomarkers for the kidney functions.⁵² As shown in Figure S15a–f, these biomarker values for both liver and kidney functions from all anti-EGFR-Gd₂O₃, anti-EGFR-Ce¹⁰B₆, and anti-EGFR-Gd¹⁰B₆ NPs-injected mice at day 14 after neutron irradiation are slightly higher than the control group, but all are lower than those for the FDA approved, clinically used chemodrug, DOX. The in vivo cytotoxicities of these three nanoparticles in healthy mice without bearing H&N tumors were also determined in a similar way (see Figure S16). All the levels of blood biomarkers corresponding to liver and kidney functions did not show any significant changes as compared to the control group. Overall, both in vitro and in vivo experiments show that both anti-EGFR-Ce¹⁰B₆ and anti-EGFR-Gd¹⁰B₆ NPs have low and acceptable levels of cytotoxicities.⁵³

Evaluation of the Recurrence Probability for H&N Tumors Using Various Biomarkers

Relapse of tumors remains as one of the major problems in the treatments of H&N and many other tumors. Many factors are strongly associated with the growths of cancer cells and cancer stem cells, and thus are tightly correlated to the probability of H&N tumor recurrence. For example, transforming growth factor-α (TGF-α) is a polypeptide growth factor that binds selectively to the epidermal growth factor receptor (EGFR). TGF-α plays an important role in the tissue regeneration and promotes tumorigenesis.⁵⁴ Similarly, overexpression of the p53 protein provides an invaluable omen for the recurrence of H&N tumors. It is well documented that elevated levels of TGF-α and p53 in H&N tumors are highly associated with tumor growth and recurrence.⁵⁵⁻⁵⁷ Cancer stem cells (CSCs, also known as tumor-initiating cells) play a key role in tumor growth, recurrence, metastasis, and chemodrug resistance.^58,59 CD44 is a specific CSCs’ membrane surface biomarker and has been involved in promotion of tumor growth and metastasis.^60,61 Prostaglandin E₂ (PGE₂) is the most abundant prostaglandin found in various human malignant tumors. Binding of PGE₂ to the EP2 and EP4 receptors on the cancer cell membrane will trigger secretion of cytokines to promote cancer cells’ growth, proliferation, agiogenesis, metastasis, as well as suppression of host antitumor immunity.⁶²⁻⁶⁶ Therefore, a high level of PGE₂ is directly associated with a high probability of tumor recurrence. In addition, hypoxia-inducible factor-α (HIF-α) is the main transcription factor determining the cellular response to hypoxia tumor domains. An elevated level of HIF-α is known to strongly correlate to tumor angiogenesis, metastasis, drug resistance, and poor patient prognosis.⁶⁷ Pimonidazole hydrochloride (PIMO) is a tumor hypoxia domain detecting fluorescence dye and can selectively accumulate in the hypoxic cancer cells.⁶⁸ PIMO is commonly used to identify the location/size of hypoxic tumor domains and thus the amount of hypoxia-inducible factor-α (HIF-α) protein. The amounts of the above biomarkers are directly proportional to the probability of tumor relapse, and thus are used in this study to evaluate the probabilities of H&N tumor recurrence after different treatment conditions.⁶⁶

At day 20 post neutron irradiation, the tumor sections of mice from various treatment groups were collected and stained with antibodies of TGF-α (antibody, ab9585), p53 (anti p53 antibody, ab131442), CD44 (anti CD44 antibody, ab157107), PGE₂ (anti PGE₂ antibody, GR10562–97) and PIMO to evaluate the levels of these tumor recurrence-related biomarkers. From all biomarker staining images, the percentages of the biomarker-positive cells were obtained by converting the corresponding antibody staining images to numerical values using Image-Pro10 software (see Figures S18–S22 for original staining images). Here, the amounts of tumor recurrence-related biomarkers from the DOX-treated group were used as a standard (i.e., treated as 100%) for comparison. As shown in Figure 5a–e), the percentages of TGF-α, p53, CD44, PGE₂, and HIF-α positive cells for (BPA-F 1 h+NR), (anti-EGFR-Gd₂O₃ NPs+NR), (anti-EGFR-Ce¹⁰B₆ NPs+NR), and (anti-EGFR-Gd¹⁰B₆ NPs+NR) groups all have a similar trend; that is, the amounts of biomarkers for the NCT-treated groups all are smaller than that for the DOX-treated group, indicating that all three NCT therapy groups have lower tumor recurrence probabilities than that for the DOX-treated group. It is worthwhile to note that the CD44 is a specific biomarker of cancer stem cells. Among all the treatment groups, the amount of CD44 for the (anti-EGFR-Gd¹⁰B₆ NPs+NR)-treated group is only 7.6% of that for the DOX-treated group and is also a quarter of that (21%) for the (anti-EGFR-Gd₂O₃+NR)/ (31%) for the (anti-EGFR-Ce¹⁰B₆ NPs+NR) treated groups. It is generally believed that cancer stem cells are responsible for the regrowth and recurrence of H&N tumors. Effective killing of cancer stem cells is very crucial for the inhibition of H&N tumor recurrence. This result unambiguously shows that combined GdBNCT is far more effective in killing cancer stem cells and thus inhibition of the H&N tumor recurrence, as compared to both chemotherapy and pure BNCT and pure GdNCT groups. To obtain a general trend, the quantities for all the above five tumor recurrence-related biomarkers were averaged and plotted in Figure S26, where the probability for H&N tumor recurrence are in the following order: DOX (100%) > BPA-F 24 h+NR (89 ± 3%) > BPA-F 1 h+NR (70 ± 2%) > anti-EGFR-Gd₂O₃ NPs+NR (49 ± 5%) > anti-EGFR-Ce¹⁰B₆ NPs+NR (41.0 ± 0.5%) ≫ anti-EGFR-Gd¹⁰B₆ NPs+NR (16 ± 1%). The results unambiguously show that (a) nanoparticles-based NCT treatment is much better than the molecular BPA-F+NR treated group; and (b) the combined GdBNCT is superior to both pure GdNCT and pure BNCT for killing the cancer stem cells as well as suppressing the probability of H&N tumor recurrence. The results are consistent with the average half-lifespan data shown in Figure 4c.

Figure 5.

The quantities of tumor recurrence-related biomarkers and staining experiments for various treatment groups, including the control group, DOX-treated group, BPA-F+NR, anti-EGFR-Ce¹⁰B₆ NPs+NR, and anti-EGFR-Gd¹⁰B₆ NPs+NR. (a)–(e) Measurement of number of positive cells for various treatment groups using (a) anti TGF-α antibody (ab9585), (b) anti p53 antibody (ab131442), (c) anti CD44 antibody (ab157107), (d) PGE2 antibody (GR10562–97), (e) hypoxia-inducible factor-α (HIF-α). (f) H&E staining of the tumor sections at a 50 μm depth of all groups after treatments. (g) Caspase 3 staining images of tumor tissues from various treatment groups at day 20 after neutron irradiation. The doses of BPA-F, anti-EGFR-Gd₂O₃ NPs, anti-EGFR-Ce¹⁰B₆ NPs, and anti-EGFR-Gd¹⁰B₆ NPs, are the same, i.e., 50 mg/kg, which is equivalent to the amounts of (36.15 mg Gd/kg, and 13.85 mg ¹⁰B/kg) for Gd¹⁰B₆, (15 mg ¹⁰B/kg) for Ce¹⁰B₆, and (43.35 mg Gd/kg) for Gd₂O₃ nanoparticles, respectively. *p < 0.05, **p < 0.01, ***p < 0.001.

The tumor sections of different treatment groups were collected for histological examination using Hematoxylin and Eosin (H&E) staining at day 20th post neutron irradiation (Figure 5f). Overall, it was observed that the mice group treated with anti-EGFR-Gd¹⁰B₆ NPs+NR showed a significantly greater number of large lesions as compared to other treatment groups. The necrotic area of tumor tissues was characterized by shallow staining, pyknosis, irregular nuclear contours, condensed chromatin, inconspicuous nucleoli, and/or pale cytoplasm.⁶⁹⁻⁷¹ These results clearly show that the therapeutic effect of GdBNCT and the extent of necrotic cellular deaths are directly related to the amount of ¹⁰B accumulation at the tumor site. Besides necrosis, the possible presence of apoptosis-induced cell deaths was also evaluated by measuring the amount of caspase proteins since overexpression of caspase proteins initiates the cell apoptosis process. Therefore, the amount of caspase proteins is a direct indication of the cell apoptosis process. The apoptotic cellular deaths of the tumor tissue were evaluated using caspase-3 staining at day 20 after neutron irradiation (see Figure 5g). The treatment group anti-EGFR-Gd¹⁰B₆ NPs+NR showed the highest extent of apoptotic cellular deaths among all treatment groups, as featured by the abundance of brown stained nuclei, while much lower noticeable apoptosis brown stained nuclei were observed for other BNCT and GdNCT treatment groups, indicating the superior cancer-cell-killing ability of combined GdBNCT therapy as compared to pure BNCT or GdNCT alone. Overall, the in vivo H&E and caspase 3 staining data showed that (a) BNCT-induced cell deaths occur mostly through the necrosis pathway, whereas the GdNCT-induced cell deaths occur mainly via the apoptosis pathway. The results are understandable because high-energy α and ⁷Li particles generated from BNCT can cause a large extent of cell membrane breakages and damages, leading to necrotic cell deaths. While the γ-ray generated from GdNCT mainly causes damages and malfunctions of cellular components, but not cell membrane breakages. Therefore, the major GdNCT-induced cell deaths occur through the apoptotic cell death pathway. (b) The combined GdBNCT-induced cell deaths occur via combined necosis and apoptosis pathways. (c) The combined GdBNCT is much more effective in killing cancer cells than the pure BNCT effect or pure GdNCT alone. Furthermore, the major organs of mice from various groups were examined and found to be not damaged by neutron irradiation (see details in Figure S24).

Cancer is one of the very important biological problems, threatening human life. According to the World Health Organization, about 10 million persons die of different cancers in 2020. The occurrence of various cancers involves many complicated biological signaling processes and mutual interactions among cancer cells/cancer stem cells as well as immune cells. Eradication of cancers is difficult and complicated, especially for the highly metastatic and recurrent glioblastoma, and head-and-neck cancers. In this work, we reported the first successful example of using combined GdBNCT to eradicate the highly metastatic and recurrent melanoma cancer. Our work unambiguously demonstrated that combined GdBNCT can be used to resolve the tough biomedical problem of curing difficult-to-cure recurrent H&N cancers.

Conclusion

In summary, we have presented a combined GdBNCT for nearly complete eradication of H&N tumors using a novel anti-EGFR-Gd¹⁰B₆ NPs (98.5% ¹⁰B enrichment) and achieved an excellent therapeutic result for eradication of recurrent H&N tumors. The combined GdBNCT-treated mice bearing H&N tumors have a very long average half-lifespan of 81 days, which is much longer than 52 d observed from pure GdNCT effect (from anti-EGFR-Gd₂O₃+NR NPs-treated group), 60 d observed from the pure BNCT effect (from the anti-EGFR-Ce¹⁰B₆ NPs+NR-treated group, 38 d from the (BPA-F 1 h+NR)-treated group, as well as the best result reported in the literature using combined radiotherapy and T cell-mediated immunotherapy for the treatment of H&N tumor-bearing mice (70 d). As evidenced by the very low values of 5 tumor recurrence-related biomarkers including TGF-α, p53, CD44, PGE₂, and HIF-α, the Gd¹⁰B₆ NPs-based combined GdBNCT therapy can very effectively cut down the probability of H&N tumor recurrence to be only 16% of that for DOX-treated mice group, and ∼1/3 relative to the recurrence probability for both pure GdNCT and BNCT treated mice groups. The excellent antitumor outcome from the anti-EGFR-Gd¹⁰B₆ + NR-treated groups is of anti-EGFR-Gd¹⁰B₆ NPs to tumor-specific targeted delivery of high doses of boron and gadolinium contents (158 μg ¹⁰B/g tumor tissue and 364.1 μg Gd/g (or 56.8 μg ¹⁵⁷Gd/g) as well as a very high tumor-to-blood (T/B) boron ratio of 4.18. Such a combined GdBNCT antitumor effect is not possible or very difficult for molecular GdNCT and BNCT drugs, due to their very low Gd and B contents, difficulty in preparing a molecular complex containing both B and Gd elements simultaneously, and the lack of tumor-targeting ability. The presence of Gd element in the nanoparticles not only provides additional GdNCT effect on destroying distant cancer cells to compensate the very short killing zone of BNCT effect but also offers the CT and MRI imaging ability to help identify the location/size of tumor as well as the biodistribution of BNCT drugs, avoiding the troublesome synthesis, storage, and adverse effects of short-half-life radioacive ¹⁸F labeling process. Both in vitro cell viability experiments and measurements of in vivo inflamatory markers of liver and kidney functions showed that the anti-EGFR-Gd¹⁰B₆ NPs have lower cytotoxicity, as compared to those of FDA approved drugs, such as BPA-F and DOX. Overall, the anti-EGFR-Gd¹⁰B₆ NPs-based combined GdBNCT not only effectively destroys the H&N tumor and prolongs the average half lifespan of mice bearing recurrent H&N tumor but also effectively kills H&N cancer stem cells and inhibits the recurrence of H&N tumor. Our results pave a new way for noninvasive clinical curing of highly recurrent H&N tumors, and possibly other recurrent tumors, in patients using combined GdBNCT therapies.

Experimental Section

Preparation and Characterization of Gd¹⁰B₆ NPs

In this study, Gd¹⁰B₆ NPs were synthesized using GdI₃ (nature abundance Gd with 15.65% ¹⁵⁷Gd), ¹⁰B-enriched boric acid (H₃¹⁰BO₃ isotropic purity ≥98.5% ¹⁰B atom%, Santa Cruz Biotechnology, Dallas, Texas, U.S.A) and magnesium (Mg) metal pieces as precursors by following a modified microwave arcing process.⁷²⁻⁷⁴ The Mg metal pieces were used to absorb microwaves and generate microwave arcing to breakdown GdI₃ and H₃¹⁰BO₃ to become atomic plasma. While cooling, the atomic fragments in the plasma will reassemble to form stable products. The morphology of as-synthesized Gd¹⁰B₆ NPs was determined using scanning electron microscopy (SEM) and transmission electron microscope (TEM). SEM images show that these particles have a rough surface morphology and a narrow size distribution with an average size of ∼50 ± 5 nm (see Figures S1a and S1b inset). A high-resolution TEM image reveals the crystalline nature of the Gd¹⁰B₆ NPs with a lattice constant of 0.410 nm (see, Figure S1b). The UV–vis-NIR absorption spectrum for Gd¹⁰B₆ NPs does not show any characteristic absorption bands in the visible regions (Figure S1f). The X-ray diffraction (XRD) pattern (Figure S1c) signifies the local crystalline nature of Gd¹⁰B₆ NPs with a set of well-defined diffraction peaks in agreement with JCPDS file index (JCPDS No. 38-1426).⁷² Gd¹⁰B₆ NPs were further characterized by using X-ray photoelectron spectrometry (XPS) to identify the chemical states of gadolinium and boron elements. As shown in the Figure 1e, the peaks at 142.7 and 148.1 eV are ascribed to Gd 4d_5/2 and Gd 4d_3/2 generated from spin orbit interaction in rare earth hexaborides (Figure S 1d).^75,76 The B 1s binding energy at 187.2 eV in Figure S1e is lower than that of other rare-earth hexaborides due to the different coordination environments. The XPS band at 191.5 eV is due to the presence of B₂O₃ impurity, which was formed in the microwave arcing process.⁷⁷ These XPS spectra match well with those of GdB₆ reported in the literature.⁷²

Preparation and Characterization of F127-COOH

Furthermore, to achieve good dispersion in aqueous solutions, Gd¹⁰B₆ NPs were coated with a layer of the F127-COOH polymer via electrostatic interactions of the carboxylate anion of F127 and Gd³⁺ cations. The F127-COOH polymer was synthesized according to a literature process.⁷⁸ In brief, Pluronic F127 (1 mmol) and 4-dimethylaminopyridine (DMAP) (1 mmol) were dispersed in 1,4-dioxane (10 mL), and then triethyl amine (60 mL) was added and mixed under an inert atmosphere. Then, succinic anhydride solution (125 mg in 5 mL of 1,4-dioxane) was added dropwise to this solution, and the reaction mixture was kept stirring overnight, followed by washing with cold dry ether. The obtained product was then allowed to dry under vacuum overnight to obtain F127-COOH. The chemical structure was confirmed by ¹³C NMR spectrum (Figure S2).⁷⁹ It is known that epidermal growth factor receptors (EGFR) are overexpressed in more than 90% of H&N tumor patients.⁸⁰ To achieve targeting ability toward H&N cancer cells, F127-Gd¹⁰B₆ NPs were further surface-functionalized with anti-EGFR antibody via 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide (EDC) coupling of the amino group in anti-EGFR moieties and the carboxylic acid group in the F127-COOH polymer chains.⁸¹ Successful conjugation was confirmed by FTIR analysis (Figure S3a). The FTIR absorption band at 1730 cm^–1 of the F127-Gd¹⁰B₆ NPs is due to the stretching vibration of the carboxyl > C=O moieties from F127 (see the red curve in Figure S3a), which was shifted to a lower energy position of 1640 cm^–1 upon conjugation with anti-EGFR, due to the keto–enol tautomerization. The FTIR bands at ∼3000 cm^–1 for both F127-Gd¹⁰B₆ NPs and anti-EGFR-F127-Gd¹⁰B₆ NPs are due to the stretching vibrations of the C–H bonds of the F127 polymers. The FTIR spectroscopic results confirm that F127-Gd¹⁰B₆ NPs were successfully modified with anti-EGFR. The net surface charges of bare Gd¹⁰B₆ NPs, F127-Gd¹⁰B₆ NPs and anti-EGFR-F127-Gd¹⁰B₆ (or shortened as anti-EGFR-Gd¹⁰B₆ NPs) in an aqueous solution were measured. As shown in Figure S3b, the zeta potential values measured at pH 7.4 were +12, −11.5, and +4.5 mV for bare Gd¹⁰B₆ NPs, F127–Gd¹⁰B₆ NPs and anti-EGFR-Gd¹⁰B₆ NPs, respectively, indicating that F127–Gd¹⁰B₆ NPs have a negatively charged surface, while bare Gd¹⁰B₆ NPs and anti-EGFR-Gd¹⁰B₆ NPs were positively charged. It was reported that in lanthanide hexaborides, it is commonly observed to have structural defects boron vacancies^82,83 especially in the presence of cubic lattice distortions, which can explain the positive surface charge of the Gd¹⁰B₆ nanoparticles. Gadolinium is a rare earth metal element with paramagnetic properties from unpaired electrons in the 4f orbitals. The magnetic susceptibility measurements of anti-EGFR-Gd¹⁰B₆ NPs were performed in the temperature range from 2 to 300 K as shown in Figure S4a. The magnetic susceptibility χ(T) continuously increases upon decreasing the temperature. A sharp increase of the magnetic susceptibility χ(T) occurs below 25 K. The magnetic hysteresis loop M(H) for Gd¹⁰B₆ NPs was measured at 2 K (Figure S4b). The magnetic susceptibility approaches a saturation value of 0.0062 emu/g at an external magnetic field of 7T. The absence of hysteresis loops indicates that Gd¹⁰B₆ NPs are superparamagnetic, that is, when the external magnetic field was removed, the net magnetic moment of Gd¹⁰B₆ nanoparticles goes back to zero; and the Gd¹⁰B₆ nanoparticles will not stack together to form large aggregates due to the absence of net magnetic moments of nanoparticles.⁸⁴

Synthesis of Anti-EGFR-Functionalized Gd¹⁰B₆ NPs

In a typical experiment, 3 mg of Gd¹⁰B₆NPs and 20 mg of F127- COOH were added to the solution at room temperature containing 2 mL of chloroform and evaporated under vacuum. In the next step, 5 mg of F127-COOH-capped Gd¹⁰B₆NPs was mixed with 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide (EDC) (3 mg, 15.8 μmol) and N-hydroxy succinimide (NHS) (2 mg, 17.4 μmol) in 1× PBS solution. To this mixture was added 20 μL (50 μg/mL) of anti-EGFR (R&D Systems, Minneapolis, USA), and the mixture was stirred for 4 h at 4 °C. The obtained anti-EGFR-functionalized Gd¹⁰B₆NPs were washed thoroughly to remove the unreacted anti-EGFR and then redispersed and stored in 1× PBS for further use.

Preparation and Characterization of Ce¹⁰B₆ NPs

Ce¹⁰B₆ NPs were prepared in a way similar to that for Gd¹⁰B⁶ NPs except using CeI₃ to replace GdI₃. SEM images show that the as-prepared Ce¹⁰B₆ NPs are homogeneous with particle sizes in the range of 50–60 nm (Figure S5a). The XPS peaks of the Ce¹⁰B₆ NPs shows sharp well-defined Ce 3d_5/2 (at 885 eV), Ce 3d_3/2 (at 905 eV), and B 1s (at 187.4 eV) peaks, which coincide with those reported in the literature (see Figure. S5c).⁸⁵⁻⁸⁷ The UV–vis-NIR spectrum of Ce¹⁰B NPs (Figure S1) did not show any characteristic absorption band, similar to that for Gd¹⁰B₆ NPs. To achieve good tumor-targeting ability, Ce¹⁰B₆ NPs were surface-modified using F127-COOH and further conjugated with anti-EGFR via 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide (EDC) coupling to form anti-EGFR-F127-Ce¹⁰B₆ (or shortened as anti-EFGR-Ce₁₀B₆) NPs. FTIR analysis was performed to confirm the successful functionalization (Figure S6a). The zeta potential values for bare Ce¹⁰B₆, F-127-Ce¹⁰B₆, and anti-EGFR-Ce¹⁰B₆ NPs were measured at pH 7.4 and found to be 14, −12.5, and 4 mV, respectively (see Figure S6b). The zeta potential values are similar to those obtained for the Gd¹⁰B₆ systems. Similar to Gd¹⁰B₆, the positive surface charge of the Ce¹⁰B₆ NPs is most probably due to the structural boron vacancies defects.^79,80 The Ce¹⁰B₆ NPs were prepared and used to exert a pure BNCT effect for comparison with the combined GdBNCT from the Gd¹⁰B₆ NPs system. By comparing the Ce¹⁰B₆ and the Gd¹⁰B₆ systems, one can obtain information about whether additional GdNCT leads to better therapeutic outcomes in the cancer treatments, as compared to the pure BNCT effect alone.

Preparation and Characterization of Gd₂O₃ NPs

Gd₂O₃ NPs was purchased from a commercial source (NRE-3020, Nano Research Elements, < 100 nm). SEM and TEM images show the commercial Gd₂O₃ NPs have a rough surface morphology with particle sizes ranging around 55–60 nm (see Figures S7a,b). The XRD peaks (Figure S7c) observed at 2θ = 28.64° correspond to the (222) plane of the cubic phase (JCPDS 3-065-3181) and those at 2θ = 33.16°, 47.64°, and 56.53° correspond to the (400), (440), and (622) planes, respectively, of the hexagonal phase.⁸⁸ To achieve good tumor-targeting ability, Gd₂O₃ NPs were surface-modified using F127-COOH and further conjugated with anti-EGFR via 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide (EDC) coupling to form anti-EGFR-F127- Gd₂O₃ NPs (or shortened as anti-EFGR- Gd₂O₃) NPs. FTIR analysis was performed to confirm the successful functionalization (Figure S8a). The zeta potential values for bare Gd₂O₃, F-127- Gd₂O₃, and anti-EGFR- Gd₂O₃NPs were measured at pH 7.4 and found to be 20, −15.5, and 5 mV, respectively (Figure S8b).

The Gd¹⁰B₆ nanoparticle has a molecular weight of 217 g/mol and contains 72.3 wt % of Gd; i.e., 1 mg of Gd¹⁰B₆ contains 0.723 mg of Gd and 0.277 mg of ¹⁰B elements. Likewise, Ce¹⁰B₆ nanoparticle has a molecular weight of 200 g/mol and contains 30.0 wt % of ¹⁰B, i.e., 1 mg of Ce¹⁰B₆ contains 0.30 mg of ¹⁰B element. In the case of Gd₂O₃ nanoparticle, the molecular weight is 362 g/mol, in which the Gd element accounts for 86.7 wt %, i.e., 1 mg of Gd₂O₃ contains 0.867 mg of Gd. The Gd element used in this report is natural abundance Gd (i.e., 15.65% ¹⁵⁷Gd) but not ¹⁵⁷Gd enriched, so that the comparison of Gd₂O₃ and Gd¹⁰B₆ is under the same common ground. If the ¹⁵⁷Gd-enriched Gd element was used, both the in vitro net cell deaths and in vivo tumor growth suppression values are expected to increase. Since the natural abundance of Gd contains 15.65% ¹⁵⁷Gd (255000 barn) and 14.8% ¹⁵⁵Gd (60900 barn) and both of them have very high thermal neutron cross sections, far higher than that (3838 barn) of ¹⁰B, the use of nanoparticles with natural abundance Gd is sufficient to fulfill the threshold Gd value of 50–200 μg ¹⁵⁷Gd/g tumor tissue for generation of effective GdNCT. Therefore, natural abundance Gd was used in this study.

To determine the thermostability of Gd₂O₃, Gd¹⁰B₆, and Ce¹⁰B₆, thermal gravity analysis measurements of Gd₂O₃-F127 COOH, Gd¹⁰B₆-F127 COOH, and Ce¹⁰B₆-F127 COOH NPs were conducted, and data are shown in the supporting Figure S9. The TGA data show that the surface-chelated F127 polymer chains (on both Gd¹⁰B₆ and Ce¹⁰B₆ NPs) are quite stable up to a decomposition temperature of ∼330 °C. In the case of Gd₂O₃ NPs, the onset decomposition of surface-chelated ligands occurs at a temperature around 60 °C, distinctly different from the onset decomposition temperature of ∼330 °C. Since the Gd₂O₃ NPs used in this study was purchased from a commercial nanomaterial company (NRE-3020, Nano Research Elements, <100 nm), we attribute the early onset decomposition behavior of Gd₂O₃ NPs to some surface-chelated small molecules from the original company. The TGA data shown in the supporting Figure S9 also show that inorganic core nanoparticles, including Gd₂O₃, Gd¹⁰B₆, and Ce¹⁰B₆ NPs are quite stable under temperatures below 450 °C.

Glossary

ABBREVIATIONS

GdNCT

gadolinium neutron capture therapy,

BNCT

boron neutron capture therapy,

GdBNCT

gadolinium–boron neutron capture therapy

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.3c00250.

• Preparation of BPA-Fructose complex, Gd leakage analysis using ICP-MS, cell culture, cell viability, EGFR targeting property using anti-EGFR-M¹⁰B₆ NPs-internalized HTB-43 cells, Calcein AM and PI staining experiments, apoptotic and necrotic Assay, Hematoxylin & Eosin (H&E) and Caspase-3 staining, T1-weighted MR imaging, Computed Tomography (CT) imaging, biodistribution of anti-EGFR-M¹⁰B₆ NPs using inductively coupled mass spectrometry (ICP-MS), gadolinium boron neutron capture therapy (GdBNCT), TGF-α, p53, CD44, PGE2 and HIF-α staining experiments for determining the probability of H&N tumor recurrence, in vivo blood biochemical analysis, as well as supporting figures, such as, characterization of the Gd¹⁰B₆ NPs, ¹³C NMR spectrum of F-127 COOH polymer, spectroscopic characterization of Gd¹⁰B₆ NPs, magnetic properties of bare Gd¹⁰B₆ NPs, characterization of the bare Ce¹⁰B₆ NPs, spectroscopic characterization of Ce¹⁰B₆ NPs, characterization of the bare Gd₂O₃ NPs, Spectroscopic characterization of Gd₂O₃ NPs, TGA analysis of Gd₂O₃–F127 COOH_, Gd¹⁰B₆–F127 COOH, and Ce¹⁰B₆–F127 COOH NPs, analysis of Gd leakage using ICP-MS, apoptosis and necrosis cellular rates, cell viabilities of a healthy human umbilical vein endothelial cells, cell viability of control cells and control cells+ (15, 30, 60 min) neutron irradiation, MTT assay for in vitro cytotoxicities of HTB-43 H&N cancer cells, biodistribution of anti-EGFR-Gd₂O₃ NPs, biodistribution of anti-EGFR-Ce¹⁰B₆ NPs, time evolution of the averaged mice body weight curves, serum biochemical analysis of mice bearing H&N tumor, serum biochemical analysis of healthy mice injected with anti-EGFR-Gd₂O₃ NPs, anti-EGFR-Gd¹⁰B₆ NPs and anti-EGFR-Ce¹⁰B₆, (50 mg/kg), in vivo pharmacokinetic studies, in vivo staining experiments for TGF-α, p53, CD44, PGE₂ and HIF-α, In vivo staining experiments, average quantities of five biomarkers, and in vivo H&E staining images of liver, spleen, kidney, lungs and heart sections. Supporting Table S1 provides an overview of various gadolinium-based nanocarriers used for GdNCT treatment, and the supporting Table S2 provides examples of treatment modality for H&N tumors in mice reported in the literature (PDF)

Author Contributions

K.C.H. conceived the idea, supervised the project, and finalized the submitted manuscript. M.S. performed the experiments with help from N.K., analyzed the data, and wrote the first draft manuscript. C.S.C. provided the animal facilities. X.K. and C.S.C. provided helpful discussions regarding the experimental designs and also helped to resolve some technical issues during experiments. CRediT: Munusamy Shanmugam data curation, formal analysis, writing-original draft; Naresh Kuthala data curation; XIANGYI KONG investigation, validation, visualization; Chi-Shiun Chiang methodology, validation, visualization; Kuo Chu Hwang conceptualization, formal analysis, funding acquisition, project administration, resources, supervision, writing-review & editing.

The authors declare no competing financial interest.