Boron Theranostic Nanoplatform Utilizing a GO@Carborane@Au Hybrid Framework for Targeted Delivery

Abstract

Background: Boron neutron capture therapy (BNCT) represents a highly selective therapeutic modality for recalcitrant cancers, leveraging the nuclear reaction initiated by thermal neutron capture in boron-10 (¹⁰B) to deliver high-linear energy transfer radiation (α-particles and ⁷Li ions) directly within tumor cell boundaries. However, the widespread clinical adoption of BNCT is critically hampered by the pharmacological challenge of achieving sufficiently high, tumor-selective intracellular ¹⁰B concentrations (20–50 μg of ¹⁰B/g tissue). Conventional small-molecule boron carriers often exhibit dose-limiting non-specificity, rapid systemic clearance, and poor cellular uptake kinetics. Methods: To overcome these delivery barriers, we synthesized and characterized a novel dual-modality nanoplatform based on highly biocompatible, functionalized graphene oxide (GO). This platform was structurally optimized via covalent conjugation with high-boron content carborane clusters (dodecacarborane derivatives) for enhanced BNCT efficacy. Crucially, the nanocarrier was further decorated with plasmonic gold nanostructures (AuNPs), endowing the system with intrinsic surface-enhanced Raman scattering (SERS) properties, enabling real-time, high-resolution intracellular tracking and quantification. Results: We evaluated the synthesized GO@Carborane@Au nanoplatforms for their stability, cytotoxicity, and internalization characteristics. Cytotoxicity assays demonstrated excellent biocompatibility against the non-malignant human keratinocyte line (HaCaT) while showing selective toxicity (upon irradiation, if tested) and high cellular uptake efficiency in the aggressive human glioblastoma tumor cell line (T98G). The integrated plasmonic component allowed for the successful, non-destructive monitoring of nanoplatform delivery and accumulation within both HaCaT and T98G cells using SERS microscopy, confirming the potential for pharmacokinetic and biodistribution studies in vivo. Conclusions: This work details the successful synthesis and preliminary in vitro validation of a unique graphene oxide-based dual-modality nanoplatform designed to address the critical delivery and monitoring challenges of BNCT. By combining highly efficient carborane delivery with an integrated photonic trace marker, this system establishes a robust paradigm for next-generation theranostic agents, significantly advancing the potential for precision, image-guided BNCT for difficult-to-treat cancers like glioblastoma.

1. Introduction

Boron neutron capture therapy (BNCT) has emerged as a promising modality within precision oncology, offering theoretical and practical advantages for treating malignancies such as glioblastoma, malignant melanoma, and recurrent squamous cell carcinomas of the head and neck as examples of tumors traditionally resistant to conventional photon-based radiotherapy and chemotherapy [1,2]. BNCT therapeutic efficacy hinges on the selective delivery of the non-toxic isotope ¹⁰B to tumor cells, followed by irradiation with low-energy thermal neutrons [3,4]. The high neutron capture cross-section of the boron triggers a nuclear reaction producing high-linear energy transfer (LET) α-particles and 7Li nuclei, which deposit lethal energy within a cellular-scale range of 5–9 μm [5]. This microscopic precision confines cytotoxicity to boron-loaded tumor cells while sparing surrounding healthy tissues and vascular structures. In addition, carboranes are used to modify targeting drugs, including nucleic acids. Kwiatkowska et al. described an approach where boron clusters were utilized to modify siRNA targeting the BACE1 gene [6], achieving interesting results. However, clinical translation of BNCT has been fundamentally constrained by the pharmacokinetics of first-generation boron carriers. Currently, only two ¹⁰B-containing compounds—¹⁰BPA (¹⁰B-phenylalanine, Stella Pharma’s Steboronine^®) and ¹⁰BSH (sodium borocaptate)—are being used in clinical trials [7,8,9]. These agents suffer from suboptimal tumor selectivity, rapid systemic clearance, and insufficient tumor-to-normal tissue ratios, compromising the requisite intratumoral boron concentration (20–50 μg/g) during neutron irradiation [10]. Addressing these critical limitations necessitates advanced drug delivery systems (DDSs) that can protect boron payloads in circulation, prolong systemic half-life, and achieve targeted release within the tumor microenvironment.

Nanomaterial-based platforms have emerged as transformative candidates for BNCT DDS, providing customizable physicochemical interfaces that enable controlled biodistribution and stimulus-responsive payload release [11]. Graphene oxide (GO) represents a particularly attractive scaffold due to its single-atom-thick, amphiphilic structure, which offers exceptional surface area and functional group diversity for high-density drug loading [12,13]. However, pristine GO is prone to aggregation in physiological media, limiting its clinical applicability. Surface functionalization with biocompatible polymers such as polyvinyl alcohol (PVA) enhances colloidal stability by introducing a steric hydration layer that mitigates serum protein adsorption and reticuloendothelial system clearance [14,15,16]. PVA-functionalized GO nanocarriers are optimally sized to evade renal filtration while taking advantage of the enhanced permeability and retention (EPR) effect for preferential tumor accumulation. Moreover, these constructs can be engineered for triggered boron release in response to acidic tumor pH or lysosomal conditions, thereby maximizing intratumoral boron delivery specificity. Graphene oxide has been previously utilized in the drug delivery of multiple compounds, including nucleic acids such as siRNA [17,18], proteins [19], and low-molecular-weight drugs [20]. GO was also functionalized using carborane for possible applications in electronics [21]. Drug delivery systems for carboranes have also been developed based on hyaluronic acid [22], where their selective interaction with CD44 was studied; iron oxide nanomaterials [21], for the evaluation of their cytotoxicity in cell lines including HeLa, BxPC-3, MCF-7, and L929; and gold nanoclusters [23], for the purpose of precise imaging of glioblastoma cells.

Building on these insights, we present a novel hierarchical nanoplatform, GO@Carb@Au, comprising a GO core functionalized with PVA, loaded with a carborane boron cluster, and decorated with gold nanoparticles. The carborane moiety significantly amplifies boron payload density beyond conventional agents, enhancing neutron capture potential. Gold nanoparticle integration provides structural reinforcement and offers additional functionality, including radiosensitization and tracing using surface-enhanced Raman microscopy, thereby facilitating future theranostic applications. Comprehensive in vitro evaluation using the T98G glioblastoma and HaCaT keratinocyte cell lines demonstrates efficient tumor-selective internalization, a pronounced cytotoxic effect in malignant cells following exposure to neutron irradiation analogs, and negligible toxicity to their healthy counterparts. These findings validate the GO@Carb@Au platform’s capability to overcome current delivery barriers and highlight its promise as a next-generation DDS tailored for BNCT. This work represents an interdisciplinary convergence of nuclear physics, nanomaterial science, polymer chemistry, and tumor biology, providing a robust strategy to redefine particle-based radiotherapeutics in oncology.

2. Materials and Methods

2.1. Preparation of GO

The initial material for the study was based on commercially produced graphene oxide (GO) obtained from Merck (Saint Louis, MO, USA). The target average flake size (<100 nm) was achieved through a combined approach involving ultrasonication, solvent, and temperature. Briefly, the GO stock solution (c = 4 mg/mL) was diluted with sterile PBS (Merck, Saint Louis, MO, USA) to 400 µg/mL. Consequently, ultrasonication was applied for 2 h at 50 °C using a Bandelin ultrasonic bath (Berlin, Germany). The sample was then agitated for 18 h at 500 RPM and 65 °C and subsequently sonicated again at 70 °C for 1 h. Larger flakes were removed by centrifugation at 21,191 RPM (revolutions per minute, Sorvall series, Fisher Scientific, Waltham, MA, USA)) for 15 min. Next, pristine GO nanomaterial was functionalized with PVA (Sigma-Aldrich, St. Louis, MI, USA) to enhance biocompatibility and increase the number of active functional groups for further anchoring of drug delivery compounds. Then, 25 mg of PVA was added to 5 mL of the previously prepared GO dispersion. The mixture underwent a 10 min sonication step. Following this, 40 µL of N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC, Sigma-Aldrich, St. Louis, MI, USA) at a concentration of 5 mg/mL was gradually added to the mixture. After 24 h of stirring (500 RPM, 65 °C), a second round of stirring and sonication followed for 18 h (500 RPM, 65 °C) and 6 h (70 °C), respectively. Infrared spectra of PVA-modified GO were acquired using a Nicolet IR spectrometer (Fisher Scientific, Waltham, MA, USA) operating in ATR mode using a ZnSe crystal.

2.2. Synthesis of GO@Carb

In total, 2 mL of the previously prepared GO@PVA flakes were ultrasonicated for 3 h. The dispersion was then mixed with 1 mL of an aqueous solution containing 100 mg/L of m-carborane-1-carboxylic acid (Carb, Sigma-Aldrich, p.a.) and 1 mL of a solution containing 1 mg/L of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, Sigma-Aldrich, St. Louis, MI, USA) and stirred at room temperature at 60 RPM for 3 h. This protocol achieved a loading of 33.3 μg of carborane on the platform at 100% loading capacity. The functionalized 2D nanomaterial was isolated by microdialysis for 18 h at room temperature, then resuspended in 2 mL of phosphate-buffered saline (pH 7.4, Merck, Saint Louis, MO, USA). Further analysis of the free carborane compound in the reaction mixture after the dialysis showed a loading capacity of 92% at the present conditions, which gave approximately 31 ug/mL concentration of carborane loaded on the platform, marking it as a possible alternative to currently applied free drugs.

2.3. Synthesis of GO@Carb@Au

The 1 mL of previously prepared GO@Carb nanomaterial was mixed with 1 mL of a PBS suspension (pH 7.4) containing gold nanoparticles (0.1 mg/mL) and shaken at 150 RPM for 6 h at room temperature. The mixture was centrifuged at 15,000 RPM at 4 °C for 30 min to separate GO from the free and considerably smaller Au nanoparticles. The separated GO was washed with PBS. The process was repeated 2 times. The final product was stored in the refrigerator at 5 °C. The overall scheme describing the synthesis process for the GO@Carb@Au is shown in Figure 1.

2.4. Characterization of the Platform Using Microscopy

The initial nanomaterials and their functionalized modalities were characterized using electron microscopy, including scanning electron microscopy (SEM, ZEISS Sigma 360, Oberkochen, Germany) and transmission electron microscopy (TEM, Delong, Brno, Czech Republic). The experimental parameters for SEM were 7–10 kV accelerating voltage; the aperture was set to 30. TEM: 5 kV accelerating voltage.

2.5. In Vitro Biological Study

2.5.1. Cell Culture

For the in vitro biological studies, two different cell lines were used: HaCaT, an immortalized human keratinocyte line, and T98G, a human glioblastoma-derived tumor cell line. Both cell types were maintained in DMEM (Gibco, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco), 1% L-glutamine (Gibco), and 1% penicillin–streptomycin solution (100 U/mL and 100 µg/mL; Gibco). All cell manipulation steps were carried out under aseptic conditions in a laminar flow hood. The cultures were incubated at 37 °C in a humidified atmosphere containing 5% CO₂. Prior to experiments, cells were detached using trypsin, pelleted by centrifugation, and resuspended in a fresh medium. Cell concentrations were determined using a Bio-Rad TC10™ automated cell counter (Marshall Scientific, Hampton, NH, USA).

2.5.2. Cell Viability (Alamar Blue) Assay

For the Alamar Blue viability assay, 8000 cells were seeded per well on a 96-well plate and allowed to adhere overnight. Cells were subsequently exposed to bare GO functionalized using PEG (labeled as “bare”), GO@Carb, and GO@Carb@Au samples at five concentrations of carborane, namely 1.75, 3.5, 7, 14, and 28 μg/mL. Prior to application, each sample was placed in an ultrasonic bath for 10 min to ensure proper dispersion and then diluted directly into a complete culture medium to obtain the desired working concentrations. The treated cells were incubated for 24 or 72 h at 37 °C under 5% CO₂. Following the exposure period, Alamar Blue reagent was added directly to the culture medium to achieve a final concentration of 10% of the total well volume. During a 6 h incubation at 37 °C and 5% CO₂, metabolically active cells reduced resazurin to fluorescent resorufin, whereas no conversion occurred in non-viable cells. Absorbance was measured using a Tecan Infinite M200 Pro plate reader (Zurich, Switzerland) at an emission wavelength of 570 nm. Cell viability was determined by comparing the mean absorbance of treated wells to that of untreated control cells. The results are expressed as percentage viability. Data are presented as mean ± standard deviation from six independent measurements.

2.5.3. Fixation of Cells for Raman Spectroscopy

Raman spectroscopy was employed as a label-free method to investigate cellular responses to the tested nanomaterials. This technique provides molecular-level insights into cellular composition and enables the detection of biochemical alterations resulting from nanoparticle uptake and intracellular interactions.

For sample preparation, sterile glass coverslips were placed into a 24-well culture plate before cell seeding. Subsequently, 30,000 HaCaT or T98 cells were added onto each coverslip in 500 μL of complete medium and allowed to adhere for 24 h. The following day, the cells were treated with GO@Carb@Au at a concentration of 1.75 µg/mL, normalized to the total carborane content. After a 24 h incubation period, the coverslips were washed three times with 1 mL of PBS to remove any remaining medium and unbound material. The cells were then fixed at room temperature for 30 min using 1 mL of a fixative solution composed of 1.5% paraformaldehyde (from a 4% stock in PBS; Thermo Fisher Scientific, Waltham, MA, USA), 0.1% glutaraldehyde (Sigma Aldrich, USA), and PBS. Following fixation, the samples were washed once with 1 mL of PBS and subsequently rinsed three times with 1 mL of distilled water to eliminate salts and minimize spectral interference. Finally, the coverslips were allowed to air-dry at room temperature and stored in a dust-free environment until Raman measurements were conducted.

2.5.4. Characterization and Study of the GO@Carb@Au Using Vibrational Spectroscopy

First, the nanomaterial was characterized using infrared spectroscopy to evaluate the effect of the carborane presence on the surface of the 2D nanoplatform on its vibrational states. The measurements were performed using a Thermo-Fisher infrared microscope using a ZnSe Attenuated Total Reflectance (ATR) crystal (Thermo-Fisher, USA). Second, the nanomaterial and fixed cells, both treated and untreated with GO@Carb@Au, were studied using Raman imaging. The Witec Alpha 300R+ Raman system (Witec G.M.B.H., Ulm, Germany) was utilized for all experiments. The laser operated at a wavelength of 532 nm (green) with a power of 5 mW on the sample. The single spectra were obtained using 64 microscans with an exposure time of 1 s. The maps were obtained by averaging four microscans, with an exposure time of 0.5 s. The quantification of the carborane concentration for the characterization of the nanoplatform and for stability studies was performed using a UV/Vis (Cary Series, Agilent Technologies, Santa Clara, CA, USA) instrument according to the protocol described by Bekbol et al. [24]. Briefly, platforms were isolated from the free drugs by centrifugation at 15.000 RPM. The free carborane in PBS was measured at 260 nm. The quantity was evaluated using a previously constructed 5-point calibration curve, measured in the range from 0.1 to 10 ug/mL, with R2 = 0.995.

2.5.5. Characterization and Study of the GO@Carb@Au Using Atomic Force Microscopy (AFM)

The as-prepared nanomaterial GO@Carb@Au was characterized using an Ntegra Spectra AFM microscope (NT-MDT, Moscow, Russia). The sample was first centrifuged at 15.000 RPM to remove the original solvent (PBS buffer), and it was exchanged for water. Then, 2 microliters of the sample containing GO@Carb@Au nanoplatform were consequently deposited on a freshly prepared HOPG (highly ordered pyrolytic graphite) and allowed to dry at room temperature. The sample was then measured in semicontact mode using CSG-30 probes. The scan speed was set to 0.3 Hz at a scan size of 512 × 512 pixels.

3. Results

3.1. Morphological and Chemical Characterization of the GO@Carb@Au Nanoplatforms

The morphological properties of the prepared GO@Carb@Au nanoplatforms were evaluated using multiple relevant microscopy techniques. Figure 2A (transmission electron microscopy, TEM) and Figure 2B (scanning electron microscopy, SEM) reveal that the nanoplatforms consist of thin, sheet-like structures characteristic of graphene oxide. The lateral size of these flakes varies, and they are clearly decorated with clusters of spherical gold nanoparticles. Atomic force microscopy (Figure 2D,E) confirmed the few-layered nature of the GO flakes and the presence of AuNP clusters on their surface. The average size of the gold nanospheres is approximately 20 nanometers, and the formed cluster islands consist of nanoparticle aggregates. This constellation indicates the formation of plasmonic hotspots with properties suitable for use as a detection system in surface-enhanced Raman spectroscopy. This potential will be evaluated in the following sections. Spectroscopic analysis confirmed the successful functionalization of the platform. The FTIR spectrum in Figure 2C displays characteristic spectral bands corresponding to the functional groups present in the composite material. Notably, peaks for ν(C=O) at 1652 cm⁻¹ and ν(C-O-C) at 1205 cm⁻¹ confirm the graphene oxide framework, while the broad ν(C-O-H) stretch found at 935 cm⁻¹ is indicative of both GO and the polyvinylacetate coating. The infrared spectrum of the functionalized GO@Carb@Au nanoplatform also contains spectral bands characteristic of carborane, including a strong band at 2550 cm⁻¹ corresponding to the ν(B-H) vibration mode. Raman imaging (Figure 2E,F) further visualizes the distribution of the carborane payload across the surface of the GO flakes using a combination of Raman imaging (in red) and optical microscopy (background image), as well as an overlapped combination of univariate analysis models showing a distribution of v(B-H) vibrations (for carborane, in red color) across the D-band of the graphene oxide (in gray-black), as shown in Figure 2F. The considerably strong signal confirms its successful attachment to the present graphene oxide flakes.

3.2. Evaluation of Plasmonic Properties

The presence of gold nanoparticles imparts plasmonic properties to the nanoplatform, making it suitable for surface-enhanced Raman spectroscopy (SERS). To demonstrate this capability, the platform was incubated for 24 h at room temperature (293K) with rhodamine 6G, a common SERS probe (R6G), at a concentration of c = 1 µM. Figure 3A shows a strong SERS spectrum of R6G obtained from four points of interest located on various flakes on the sample. Spectral data are labeled in orange, red, blue, and violet trace colors, indicating robust signal enhancement of the analyte at all places, albeit at moderately different signal intensities. This effect could originate from the variations in the distribution of gold nanoparticle islands on the surface and from the presence of available hotspots. The signal of the GO, represented, e.g., by the D and G bands at 1250 and 1590 cm⁻¹, varies reliably with the structure of the flakes and their thickness. The green signal shows the Raman signal obtained upon incubating rhodamine 6G at 1 mM with pristine GO. The spectral data indicate a considerable drop in the intensity of R6G-specific bands compared to intensities obtained using a material with the presence of gold nanoparticle islands on its surface. The signal amplification factor, calculated as the ratio of amplified to non-amplified signals measured at defined concentrations, is, on average, 1 × 10⁴. The Raman map in Figure 3B visualizes the distribution of the SERS signal across a single GO flake, revealing that the enhancement is concentrated in “hotspots” localized at previously identified AuNP clusters.

3.3. Confirmation of Carborane Functionalization

Raman spectroscopy provided direct evidence for the successful conjugation of carborane. Figure 4A shows the Raman spectrum of pure m-carborane-1-carboxylic acid, which is dominated by a set of sharp and intense spectral bands at approximately 2600 cm⁻¹, corresponding to the v(B-H) stretching mode. The SERS image in Figure 4D maps the intensity of this v(B-H) signal in red-blue grading from light red to dark blue colors, illustrating the distribution of carborane across multiple flakes (shown in the shades of gray), with the carboranes covering a majority of the surface (shown in red color) and concentrated in the center of the flakes (shown in dark blue). Figure 4B shows a SERS signal of the nanoplatform surface averaged over the center of the flake shown in Figure 4D (labeled as Spot 1) and an average spectrum taken from the edge of the flake (labeled as Spot 2). The resulting average Raman intensity of carborane-specific spectral bands from various spots on the nanoplatform indicates that the variations in the presence of functional groups binding carborane on the edges and in the center of the flake play a role in its distribution patterns. Figure 4C compares the spectra of the bare nanoplatform and the final GO@Carb@Au product. The characteristic carborane peak is clearly present in the spectrum of the final product and absent from the bare platform, unequivocally confirming successful functionalization. The stability of the loaded nanoplatform was evaluated under storage conditions (280K) over a 10-day period. The data in Figure 4E indicates that the material is stable over this period, with 10% of the drug released during this time frame. Next, the release study was performed in PBS buffer at neutral (pH 7.4) and acidic (pH 5.4) conditions over a 72 h time frame. The data shown in Figure 4F indicate a preferential release of the loading under acidic conditions, which is well aligned with the PVA-functionalized graphene oxide and the strong binding between the drug and the platform.

3.4. Evaluation of the Cytotoxicity

The biocompatibility of nanoplatforms is crucial for any potential clinical application. Figure 5 shows the results of cytotoxicity assays performed on T98 glioblastoma cells and HaCaT keratinocytes after 24 and 72 h of incubation. Across all tested concentrations (1.75 to 28 µg/mL), the bare, carborane-functionalized, and gold-decorated platforms exhibited minimal toxicity. Cell viability remained high (generally above 80–90%) for both cell lines at both time points. This excellent biocompatibility suggests that the nanoplatform is a safe vehicle for cellular delivery.

3.5. Studies of Cellular Internalization

To be effective, the boron delivery agent must be internalized by the target cancer cells. Despite the existence of other fluorescence-based microscopies, we utilized SERS mapping to visualize the uptake of the GO@Carb@Au nanoplatform. This approach does not require additional fluorescence labeling with unstable or photobleaching structures and thus offers a clearer and less biased image. Figure 6 displays the imaging results for T98 and HaCaT cells. The SERS maps (Figure 6A,C) and the overlapped optical/SERS images (Figure 6B,D) clearly show that the nanoplatforms are located within the cellular cytoplasm. SERS maps were constructed using a univariate analysis targeted at graphene oxide (D-band, in orange color), and the optical overlays contain the SERS map of graphene oxide as a D-band (in orange color) and carborane as a v(B-H) band (in blue color) against an optical microscopy background. It can be seen that the carborane is localized on the surface and also, importantly, around the edges of the oxide graphene flakes, suggesting its release into the cytoplasm. The corresponding average spectra (C) differentiate between the cell’s intrinsic signal (labeled as Spot-1) and the platform’s signal (labeled as Spot-2). Spectra from Spot-1 (cytoplasm) show broad bands characteristic of proteins and lipids, while spectra from Spot-2 (internalized platform) exhibit the distinct D and G bands of graphene oxide, together with a weaker, albeit still present, signal of carborane at 2550 cm⁻¹, confirming the intracellular presence of the nanoplatform. Interestingly, the overall amount of GO@Carb@Au is considerably lower in HaCaT cells. The spectral analysis of data shown in Figure 6B,E uncovered a > 50% difference, which indicates a considerable level of selection of the carbon-based platform to internalize preferentially in cancer cell lines.

4. Discussion

The primary objective of this study was to develop and characterize a multi-functional nanoplatform capable of delivering carborane for BNCT applications while offering a secondary modality for SERS-based tracking. The results confirm the successful synthesis of a GO@Carb@Au platform that meets these criteria. The combination of electron and atomic force microscopy provided a comprehensive morphological characterization, confirming a structure well-suited for drug delivery: high-surface-area GO flakes decorated with plasmonically active gold nanoparticles. The successful attachment of carborane, the boron-containing payload, was the most critical step. While FTIR suggested successful surface-chemistry modifications, Raman spectroscopy provided additional evidence. The distinct ν(B-H) vibrational band of carborane served as a unique spectral fingerprint, and its presence in the final product (Figure 4) confirmed that the carborane was successfully conjugated to the GO surface. This demonstrates the power of Raman spectroscopy as a non-destructive quality-control tool for verifying the functionalization of these nanocarriers. A key innovation of this platform is its theranostic design. The gold nanoparticles are not merely structural; they provide SERS functionality, as demonstrated with R6G (Figure 3). This built-in optical reporting capability was then leveraged to track the platform’s fate in a biological environment. The SERS imaging experiments (Figure 6) conclusively showed that the nanoplatforms were efficiently internalized by both cancerous (T98) and non-cancerous (HaCaT) cells, localizing within the cytoplasm. From the perspective of the involved mechanism, there is emerging evidence that boron clusters per se can function as selectors for EGFR-driven uptake without being linked to frequently used ligands (peptides, antibodies, etc.). In EGFR-overexpressing cells, including glioblastoma, probes bearing single carborane or metallacarborane cages targeted present EGFR subdomains, indicating that the receptor-dependent internalization of the clusters was probably initiated by the cage–receptor interaction [25]. The mechanism could be that the hydrophobic/3D-aromatic and, for metallacarboranes, the anionic surface of the cage engages complementary patches on the EGFR ectodomain to produce low-affinity contacts. When multivalently displayed, these contacts are sufficient to recruit clathrin-mediated endocytosis and endo-lysosomal trafficking typical of activated EGFR [26]. However, current data suggest that such ligand-free recognition is modest and context-dependent (cluster type, charge, valency, and presentation). The approach described here practically relies on multivalency or nano-assembly to reach BNCT-relevant boron loading and selectivity rather than on high-affinity monovalent binding [27,28]. Furthermore, the platform demonstrated excellent biocompatibility. The low cytotoxicity observed in both cell lines, even after 72 h of exposure (Figure 5), is a prerequisite for any material intended for in vivo use. This suggests that the GO@Carb@Au construct is a safe and stable delivery vehicle.

5. Conclusions

In this work, we have synthesized and characterized a novel GO@Carb@Au theranostic nanoplatform. Multi-modal characterization confirmed the successful decoration of graphene oxide flakes with gold nanoparticles and the covalent attachment of carborane. The platform exhibits strong SERS activity, enabling its label-free tracking once internalized by cells. Importantly, the nanoplatform shows high biocompatibility and is readily taken up by both glioma and keratinocyte cells, with a stronger preference for internalization in T98 cells, resulting in cytoplasmic accumulation. These findings highlight the significant potential of this hybrid nanomaterial as a safe and effective delivery system for boron neutron capture therapy, with integrated optical tracking capabilities that could aid in future diagnostics and treatment monitoring.