Ferrofluids Based on Anionic Polysaccharide-Coated Magnetic Nanoparticles for Targeted Magnetocatalytic-Driven Multimodal Anticancer Therapy

Abstract

Regrettably, glioblastoma multiforme (GBM) remains the deadliest form of brain cancer, with a very unfavorable prognosis for life expectancy for the patient. We report, for the first time, the green colloidal synthesis of cobalt-doped magnetic iron oxide nanoparticles (Co-MNPs) as aqueous ferrofluids, using two anionic polysaccharide biopolymers, hyaluronic acid (HA) and carboxymethyl cellulose (CMC), as surfactants. These ferrofluids based on magnetite nanoparticles (HA@Co-MNP and CMC@Co-MNP) demonstrated superparamagnetic properties and magnetic-to-thermal conversion upon exposure to an alternating magnetic field (AMF), with the extent of conversion dependent on surfactant type. In addition, the ferrophase acted as a nanozyme, mimicking peroxidase-like activity in response to hydrogen peroxide, which is present at higher levels in tumor cells. The coupling of magnetic-heat capabilities with biocatalytic behavior enhances glioblastoma cell elimination and suppresses 3D neurospheroid growth. The results also showed that active targeting based on the HA biopolymer shell, due to its affinity for CD44 membrane receptors overexpressed in GBM, outperformed CMC-coated ferrofluid analogs. These magnetocatalytic-responsive nanoplatforms offer a broad avenue for the diagnosis and therapy of numerous cancers, potentially improving patients’ quality of life and prognoses.

1. Introduction

Primary brain cancers, particularly glioblastoma multiforme (GBM), remain among the most lethal and therapeutically challenging malignancies, presenting a very unfavorable prognosis for patient life expectancy [1]. Despite the undeniable progress achieved in the last 2–3 decades in the treatment involving maximal surgical resection, followed by radiotherapy and chemotherapy, the median survival of GBM patients rarely exceeds 12–16 months [1,2]. Consequently, there is a critical gap in the development of innovative therapeutic platforms capable of selective targeting and integrated multimodal action to overcome tumor resistance while reducing the toxicity to healthy tissue/cells [1,3].

Nanotechnology, particularly nanomedicine in combination with biomedical fields, has emerged as a frontier for the design of nanosystems for the therapy of numerous cancers. In this context, iron oxide-based nanoparticles (MNPs) are of particular interest due to their superparamagnetic properties, which enable magnetic-to-thermal conversion upon exposure to an exogenous alternating magnetic field (AMF). Also, they behave as nanozymes (i.e., enzyme-like nanomaterials), mimicking peroxidase-like (i.e., POD-like) activity to catalyze the formation of reactive oxygen species (ROS) from endogenous hydrogen peroxide, which is usually present at higher levels in tumor cells than in healthy cells [4,5,6,7].

In this sense, the coupling of this magnetic-to-heat-induced capability with POD-like biocatalytic behavior offers the fascinating opportunities of novel multimodal therapy that targets the tumor through both magnetic hyperthermia (MHT) and chemodynamic (CDT) pathways, where the responses could also be boosted by doping the nanosized ferrites with transition metals such as cobalt [8,9]. Nevertheless, the literature still lacks scalable nanoplatforms that integrate tumor targeting and therapy within a single integrated biocompatible system capable of bypassing brain biological barriers, reaching the tumor site, and effectively eliminating resistant cancer cells without inducing severe systemic toxicity [1,3].

A scalable option is the use of ferrofluids, which offer numerous advanced properties for nanomedicine applications and are suitable for developing large-scale manufacturing processes, particularly chemical bottom-up approaches such as co-precipitation, which offer additional advantages in cost-effectiveness and reproducibility [10]. However, the colloidal stability and biocompatibility of these single-domain magnetic nanoparticles, even in the presence of AMF and in high-ionic physiological environments, depend on the surface coating, due to their very small dimensions [11,12,13]. Considering their potential as nanomedicines for oncological applications, biopolymers such as hyaluronic acid (HA) and carboxymethyl cellulose (CMC) are excellent candidates for chemical stabilization of nanostructures, serving as anionic surfactants for the green colloidal synthesis of aqueous ferrofluids. Both CMC and HA can create physical (i.e., steric hindrance) and electrical (i.e., electrostatic repulsion of negatively charged groups) barriers to promote the colloidal stability of the ferrophase [14]. Moreover, they provide functional groups for further conjugation with targeting molecules. It is also noteworthy that, in addition to these advantages, HA has been reported to target GBM cells and facilitate receptor-mediated transport across the BBB (blood–brain barrier) [15,16]. These features are associated with its high affinity for CD44 membrane receptors, which are highly overexpressed in glioblastoma cells and in a subpopulation of cancer stem cells (CSCs) that are extremely invasive and metastatic [17].

Hence, in this research, the synthesis and extensive physicochemical, magnetic, catalytic, and biological characterization of aqueous ferrofluids based on superparamagnetic cobalt-doped magnetite nanoparticles chemically stabilized with two anionic polysaccharides, CMC and HA, for targeted magnetocatalytic-driven multimodal anticancer therapy are reported.

2. Materials and Methods

2.1. Materials

Two polysaccharides were selected as anionic surfactants for stabilization of ferrofluids: (I) sodium carboxymethylcellulose (CMC) with degree of substitution, DS, 0.7, molar mass, MM, 90,000 Da (catalog number #419273, Sigma-Aldrich, St. Louis, MO, USA); and (II) hyaluronic acid sodium salt (HA) with MM = 80,000–100,000 Da (catalog number #FH63427, Biosynth Carbosynth Group, Compton, UK). To avoid redundancy with published literature, common reagents and conventional materials used in this work are listed in the Supplementary Material.

2.2. Ferrofluid Synthesis

A co-precipitation approach through “bottom-up” colloidal chemistry was employed to synthesize cobalt-doped magnetic nanoparticles (Co-MNP), as outlined in the Supplementary Material. Cobalt doping, i.e., a well-controlled partial substitution of the divalent transition-metal cation endowing new characteristics to the hosting nanosystem, replaced 40 mol% of Fe²⁺ with Co²⁺ species (Co_xFe_3−xO₄, x = 0.4, Co_0.4Fe_2.6O₄). This value corresponds to a theoretical molar content of 5.7 mol% (or 10.9 wt%) of cobalt in the ferrite, which increases magnetic and catalytic responses of the nanoferrites, while presenting limited cytotoxicity for biomedical applications.

The concentration of the polysaccharides were 1.0% w/v and 1.2% w/v of CMC and HA, respectively (i.e., 1.1 ± 0.1%; 1.0% vs. 1.2% w/v), corresponding to approximately Metal:Carboxylate of 1:0.5 mol:mol. Fixing the ratio between carboxylate groups and iron (or transition metal) groups is pivotal in bottom-up nanoparticle synthesis, where colloidal systems are produced through well-controlled nucleation, growth, and stabilization stages, and the proportion Metal (M):Ligand (anionic polymer, Lpol) is critical (i.e., M^x+: Lpol^y− → kinetics and thermodynamics).

In summary, Co²⁺, Fe²⁺, and Fe³⁺ ions were dissolved and homogeneized in the polysaccharide solutions in a ratio of 0.02 M of Fe²⁺ + Co²⁺ and 0.04 M of Fe³⁺. In the sequence, under a nitrogen atmosphere, the medium was heated to 80 ± 2 °C, and the pH was raised to alkaline levels using ammonium hydroxide solution. The obtained ferrofluids were referred to as HA@Co-MNP and CMC@Co-MNP depending on the polysaccharide surfactant used, hyaluronic acid or carboxymethylcellulose, respectively.

2.3. Polymer and Ferrofluid Characterization

The characterization of polysaccharides and ferrofluids was carried out relying on Fourier transform infrared spectroscopy (FTIR, transmission mode, Nicolet 6700, Thermo Fisher Scientific Inc., Waltham, MA, USA) and, for surface analysis, X-ray photoelectron spectroscopy (XPS, Mg-Kα radiation, 120 W, Amicus spectrometer, Kratos, Kawasaki, Kanagawa, Japan). Samples were films dried from polymer solutions and ferrofluid suspensions. For ferrofluid suspensions, zeta potential (ZP) and dynamic light scattering (DLS) analyses were performed using the ZetaPlus system (Brookhaven Instruments Corporation, Holtsville, NY, USA).

The spectroscopic, structural, morphological, and magnetic features of the inorganic core were also assessed using transmission electron microscopy (TEM, 200 kV, TEM, Tecnai G2-20, FEI Company, Hillsboro, OR, USA) coupled with energy-dispersive X-ray spectroscopy (EDX, Xplore, Oxford Instruments, Abingdon, UK), X-ray diffraction (XRD, Cu-Kα radiation, PANalytical Empyrean diffractometer, Malvern Panalytical Ltd., Malvern, UK), X-ray fluorescence spectrometry (WD-XRF, Supermini200 spectrometer, Rigaku Corporation, Tokyo, Japan), electron paramagnetic resonance (EPR, 77 K and 300 K, MiniScope MS400, Magnettech, Freiberg, Germany), and vibrating-sample magnetometer (VSM, 300 K, 7404 VSM, Lake Shore Cryotronics, Inc., Westerville, OH, USA).

The evaluation of magnetic-to-thermal conversion capacity was performed using the Magnetherm™ instrument (nanoTherics, Staffordshire, UK) at different magnetic field strengths (H = 10, 15, 20 kA/m) and a frequency of f = 110 kHz. The AMF used in the experiment is water-cooled (T_water ~24 °C), and the plastic vial (cryotube) with ferrofluid was positioned within the coil using an insulating (expanded polystyrene) sample holder, which maintains the vial sample in the region of maximum field homogeneity. The heating procedures were extended to 60 min to allow all samples to reach steady-state thermal equilibrium (the thermal energy generated by the MNP is equilibrated with heat losses to the surroundings) across all experimental conditions and to evaluate whether the nanomaterials remain stable following a long AMF stimulation period. The temperature changes in the samples were measured with a non-ferromagnetic Type T thermocouple (Copper/Constantan) with twisted-wire geometry to reduce direct interaction with the magnetic field [18,19,20]. The specific absorption rate (SAR) was calculated using the initial slope method, as described in Equation (1) [21,22].

$$\mathrm{S}\mathrm{A}\mathrm{R}={\displaystyle \frac{\mathrm{c}\mathsf{\rho }}{\mathsf{\varphi }}}{\displaystyle \frac{\mathsf{\Delta }\mathrm{T}}{\mathsf{\Delta }\mathrm{t}}}$$

(1)

where c is the specific heat capacity of the colloid (J/g·°C), ρ is the colloid density (g/mL), ϕ is the metal concentration (Fe + Co, g/mL), and ΔT/Δt is the experimentally measured heating rate (°C/s).

Aiming to evaluate heat losses, Box-Lucas fits were also performed (Equation (2)) to estimate the linear loss parameter, L (Equation (3)), for HA- and CMC-stabilized ferrofluids, and SAR_Box-Lucas was calculated using the fitting parameters (Equation (4)) [23,24].

$$∆\mathrm{T}=\mathrm{a}(1-{\mathrm{e}}^{-\mathrm{b}\left(\mathrm{t}-{\mathrm{t}}_0\right)})$$

(2)

$$\mathrm{L}=\mathrm{b}·\mathrm{C}$$

(3)

$${\mathrm{S}\mathrm{A}\mathrm{R}}_{\mathrm{B}\mathrm{o}\mathrm{x}-\mathrm{L}\mathrm{u}\mathrm{c}\mathrm{a}\mathrm{s}}=\mathrm{a}·\mathrm{b}·\mathrm{C}/{\mathrm{m}}_{\mathrm{M}\mathrm{N}\mathrm{P}}$$

(4)

where C is the heat capacity of the colloid (J/K), a (K) and b (1/s) are fit constants, L is the linear loss parameter (W/K), m_MNP is the mass of metal in the nanoparticles (g), and (t − t₀) is the experimental time (s).

The study of the catalytic properties was performed at room temperature (RT, 25.0 ± 2.5 °C) based on the oxidation of 3,3′,5,5′-tetramethylbenzidine hydrochloride (TMB) by the H₂O₂ substrate mediated by the ferrofluids (i.e., acting as nanozymes) at pH 5.0 (phosphate-citrate buffer), as previously described by our group [7]. TMBox species were detected at wavelength 654 nm using a microplate (96-well) absorbance reader (iMark^™, Bio-Rad Laboratories, Inc., Hercules, CA, USA). The Michaelis–Menten (M–M) kinetic model curves (Equation (5)) were obtained by plotting the respective initial velocities (v₀ calculated using TMBox molar extinction coefficient, ε = 3.9 × 10⁴ M⁻¹·cm⁻¹ [25]) versus H₂O₂ substrate concentration ([S]) and the maximal velocities (V_max) and Michaelis constant (K_m) were calculated using the Lineweaver–Burk plot in Origin software (OriginPro, v.2015 SR2, OriginLab Corporation, Northampton, MA, USA). Electron paramagnetic resonance tests, combined with the spin-trap methodology using DMPO (5,5-dimethyl-1-pyrroline n-oxide), were also performed to confirm the formation of ROS, as previously described by our group [26].

$${\mathrm{v}}_0={\displaystyle \frac{{\mathrm{V}}_{\mathrm{m}\mathrm{a}\mathrm{x}}\times \left[\mathrm{S}\right]}{{\mathrm{K}}_{\mathrm{m}}+\left[\mathrm{S}\right]}}$$

(5)

2.4. Cellular Experiments

In vitro tests were performed using U-87 MG (referred to as U87, American Type Culture Collection, ATCC^® HTB-14^™, purchased from the Banco de Células do Rio de Janeiro, Brazil) and HEK 293T (referred to as HEK, ATCC^® CRL-1573^™, supplied by the Federal University of Minas Gerais, Belo Horizonte, MG, Brazil). These cell lines were cultured in DMEM (with 10% FBS) at 37 °C with 5% CO₂.

Biocatalytic activity of ferrofluids in vitro was evaluated as reported in the literature [7] by the direct measurement of ROS based on the fluorescent probe DCF-DA (2′,7′-dichlorodihydrofluorescein diacetate). Also, lipid damage due to ROS was assessed by measuring MDA levels based on the thiobarbituric acid method (TBAR) [26].

To evaluate the magnetocatalytic therapeutic potential of ferrofluids in a 2D cell culture model, the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) bioassay was performed, which is the most widely accepted test for assessing the cytotoxicity of (nano)materials for biomedical applications, as previously described by our group [22]. For ROS-based therapy (ROS+), HA@Co-MNP and CMC@Co-MNP were incubated with cells at concentrations ranging from 0.003 to 100 µg/mL (corresponding to the inorganic core concentration) for 24 h to determine the half-maximal effective concentration (EC₅₀). For magnetocatalytic therapy, cells were incubated for 24 h at 0.6 μg/mL. After that, the hyperthermia groups (AMF+) were exposed to an alternating magnetic field (20 kA/m, 110 kHz) for 60 min, followed by the MTT test to evaluate cell viability compared with the group without MHT.

For 3D cell culture tests, neurospheroids were grown for 10 days as previously described by our group [27]. Then, they were treated (Day 0, 1st dose) with HA@Co-MNP and CMC@Co-MNP (0.6 and 6.0 µg/mL). After 3 days, a second dose of ferrofluids was administered (Day 3, 2nd dose). On Day 7, cell viability was assessed using the MTT assay [27]. For the AMF+ group, neurospheoids were exposed to AMF (20 kA/m and 110 kHz for 60 min) before the MTT assay. Control samples were untreated neurospheroids (ROS− and AMF−) and neurospheroids without ferrofluid treatment but exposed to AMF on Day 7 (ROS−/AMF+). Bright-field images of spheroids were captured at Day 0 and Day 7, before and after AMF exposure (Eclipse Ti-U microscope, Nikon Instruments, Melville, NY, USA).

3. Results and Discussion

3.1. Colloidal Characterization of Ferrofluids Stabilized by HA and CMC Anionic Polysaccharides

It is of paramount importance to characterize ferrofluids using a colloidal chemistry approach, in which the anionic biopolymers HA and CMC play a pivotal role in the nucleation (N), growth (G), and stabilization (S) of the inorganic Co-MNP nanoparticles in aqueous media.

In this view, CMC is a linear cellulose derivative made by D-glucose monosaccharides linked by β-1,4-glycosidic bonds, where carboxymethyl groups partially replace (depending on the degree of substitution) OH groups at positions O-2, O-3, or O-6 (Figure 1A), rendering hydroxyl and carboxyl functional groups [28]. On the other side, HA (Figure 1B) is also an unbranched polysaccharide having a repeating unit of a disaccharide of glucuronic acid and N-acetyl-D-glucosamine that are linked by alternating β1–4 and β1–3 glycosidic linkages [29]. The glucuronic acid (GlcA) unit is derived from glucose oxidation, where the primary alcohol group at C6 was converted to a –COOH group [30]. N-acetyl-D-Glucosamine (GlcNAc) is also a monosaccharide derivative of glucose, in which the hydroxyl group at C2 was replaced with an N-acetylamino group [31]. This disaccharide unit gives rise to the hydroxyl, carboxyl, and acetamido functional groups of the HA polymer.

FTIR spectra revealed these main functional groups of anionic polysaccharides and the carbon backbone, as depicted in

Table 1. FTIR vibrational band assignments of HA and CMC polysaccharides [32,33,34,35].

Assignment	Wavenumber (cm−1)
	HA	CMC
νOH (H–bonded)	3500–3100	3500–3100
νN-H (H–bonded)	3150–3050	-
νC–H	3000–2800	3000–2800
νasCOO−	1650 and 1610	1650 and 1590
νsCOO−	1420 and 1325	1410 and 1325
Amide I	1650	-
Amide II	1560	-
δC–H	1380	1380
νC–OH primary alcohol	1080	1024 and 995
νC–OH secondary alcohol	1046	1100 and 1060
νC–O–C	1152	1155

and shown in Figure 1C. The spectra confirmed the similarity of the functionalities of these polysaccharides except for the presence of the acetamido group (–N(H)–C(=O)) in the GlcNAc unit.

For both polymers, the region below 950 cm⁻¹ includes complex overlapping of deformations and stretching [32,33]. However, upon Co-MNP synthesis (Figure 1D,E), the stretching bands of the ferrophase (Fe–O and Co–O) were found at frequencies lower than 650 cm⁻¹ (615–600 cm⁻¹ for the octahedral group of Co-O and 590 cm⁻¹ for the tetrahedral group Fe–O), overlapped with these vibrations of polymeric surfactants [22,27,33]. In addition, the polysaccharide (HA and CMC) binding to the ferrophase surface resulted in several alterations in the IR spectra, including wavenumber shifts (mainly, ν_asCOO⁻ band at 1600–1610 cm⁻¹), changes in the inter-/intrachains hydrogen bonds (3600–3000 cm⁻¹), some protonation of CMC (1730 and 1235 cm⁻¹), and changes in the relative intensity of specific vibrational bands [33].

For surface chemical characterization, high-resolution X-ray photoelectron spectroscopy (HR-XPS) was performed, focusing on C, O, and N species (Figure 2A,B). For both polysaccharides, in C 1s spectra, deconvolution of curves revealed four major sub-peaks, which can be attributed to C–C/C–H bonds, C–O/C–OH bonds, N–C=O/O–C–O bonds, and O=C–O bonds. Also, the O 1s spectra showed sub-peaks of R–C=O, C–OH/C–O, and O–C=O. For HA, the signal of the amide linkage peak (N–C=O) was detected [34,35,36,37,38]. Upon ferrofluid synthesis, narrow-energy-range analysis revealed changes in the intensities of the characteristic binding energies (BE) of neat polysaccharides. Also, for some peaks, shifts in BE values were observed.

DLS analysis revealed hydrodynamic diameters (DH) consistent with an aqueous colloidal supramolecular hybrid ferrofluid, composed of the magnetic inorganic core and the polysaccharide shell, where HA-based presented a higher hydrodynamic size (ca. 116 nm) than that of the CMC-stabilized suspensions (ca. 91 nm). Zeta potential (ZP) surface charges are consistent with the presence of anionic functional groups in the polysaccharides (−63 mV and −71 mV for HA and CMC surfactants) and indicate ferrofluid stabilization as a result of the combination of electrostatic repulsion of adjacent charged groups with steric hindrance of polymer chains (Figure 2C).

3.2. Characterization of Ferrophase Stabilized by HA and CMC Polysaccharides

TEM images (Figure 3(Aa,Ba)) indicated that both ferrofluids were made of an inorganic core (ferrophase) with an assessed typical mean diameter of 5–6 nm (Figure 3(Ab,Bb)) and displayed a predominant spherical morphology. The average nanoparticle size is critical for the application, indicating that Néel relaxations are expected to be predominantly active in magneto-to-thermal conversion [39,40]. A slight contrast was detected at the nanoparticle surfaces, ascribed to a “dry” polysaccharide shell layer overcoating the ferrophase.

High-resolution (HR) TEM images, for both surfactants, revealed the presence of interference fringes compatible with the interplanar distances (d ± 0.1 Å) of the nanosized magnetite (i.e., nanomagnetites) according to the JCPDS (Joint Committee on Powder Diffraction Standards)-89-0691 file (Figure 3(Ac,Bc)). This crystalline cubic spinel structure of ferrites (and their doped derivatives) was also observed in SAED (selected-area electron diffraction) analysis (Figure 3(Ad,Bd)) and confirmed by XRD (Figure 3(Ae,Be)). It is noteworthy that the co-precipitation route used in this work was specifically designed to produce single-phase magnetite (e.g., salt precursors, concentrations, Fe²⁺(Co²⁺):Fe³⁺ 1:2 mol:mol ratio, controlled pH, N₂ non-oxidant environment, thermodynamics, kinetics, etc.), as Co_0.4Fe_2.6O_4, thereby avoiding multiphase formation. In addition, based on the (311) plane of the XRD pattern, crystallite sizes were calculated using Scherrer’s equation to be about 6–7 nm for both HA and CMC surfactants. Compared with diameters measured from microscopy images, these values indicate that the nanoparticles correspond mostly to monocrystals.

In chemical elemental analysis, EDX spectra showed the incorporation of Co species into the magnetite nanoparticle structure as a metal dopant, and peaks associated with the chemical constituents of biopolymers (HA and CMC) were observed (Figure 3(Af,Bf)). Cobalt (Co) and iron (Fe) content were estimated using WD-XRF. The results indicated Co/Fe mass ratios of 0.16 ± 0.01 for CMC@Co-MNP and 0.18 ± 0.01 for HA@Co-MNP, and the theoretical value was 0.16 [molar equivalent, Co_0.4Fe_2.6O₄, 0.4/2.6 = 0.15]. Based on these results, cobalt cations were effectively incorporated by partial substitution of Fe²⁺ ions, as designed, within the experimental statistical variation for very small nanoparticles (i.e., <13% HA@Co-MNP).

3.3. Characterization of Magnetic Properties of Ferrofluids

The RT (300 K) hysteresis curves of magnetization versus the external magnetic field obtained from colloidal suspensions are shown in Figure 4A (HA) and Figure 4B (CMC). Upon application of an external magnetic field (AMF), the systems align in the field direction, and the VSM curves exhibit similar behavior for cobalt-doped magnetite ferrophase stabilized by the anionic polysaccharides HA and CMC. The magnetization curves indicate superparamagnetic behavior for both systems, with saturation, zero remanence, and zero coercivity, consistent with the magnetite-based core size [14,41,42,43,44].

This superparamagnetic property was confirmed by EPR analysis (Figure 4C,D), which also showed the characteristic broadening associated with Co²⁺ partial substitution in magnetite (Co-doped ferrite) and revealed that the ferrofluids maintained this behavior at 77 K (Figure 4E,F).

The magneto-to-thermal conversion capability of ferrofluids was demonstrated by an increase in temperature over time upon exposure to an AMF. In response to an AMF, heat is expected to be generated from the superparamagnetic iron oxide nanoparticles primarily through Néel relaxations due to the size of the nanoparticles (~5 nm, below 12 nm) as schematically depicted in Figure 5A [39]. The AMF will essentially induce the magnetic moments to align with the applied field (spin rotation), and the Néel mechanism is characterized by heat dissipation from electromagnetic energy losses due to the reorientation of magnetic moments over an energy barrier within the magnetic core [41,45,46,47].

First, the heating efficacy of HA@Co-MNP was investigated under magnetic fields of different powers (H = 10, 15, and 20 kA/m), with the frequency fixed at 110 kHz. As expected, a large amount of heat was generated by increasing the intensity of the magnetic field (Figure 5B). Specific absorption rate (SAR), an essential parameter for evaluating the magnetic-to-heat conversion ability of magnetic materials, which describes the heat generated per second per gram of magnetic element (i.e., Fe + Co), followed the same trend (Figure 5C, left axis) [48,49].

Based on these results, H of 20 kA/m, f of 110 kHz (H × f = 2.2 × 10⁹ A/m.s < 5 × 10⁹ A/m·s, Hergt’s limit), were used in the experiments, as they fulfill the clinical safety recommendations/limitations for H, f, and H × f to avoid damage to the body/healthy cells due to induced eddy currents and peripheral muscle stimulation [48,50,51].

When comparing the effect of polysaccharide (H = 20 kA/m and f = 110 kHz, Co-MNP = 2.5 mg/mL, Fe + Co = 1.81 mg/mL), a higher temperature increase (~34%, Figure 5D,E, right y-axis) and associated SAR (~25%, Figure 5E, left y-axis) values were observed for the HA-stabilized nanoparticle. Regarding the stability of ferrofluids under these AMF exposure conditions, both ferrofluids remained chemically stable, without forming aggregates or precipitates, and retained their original hydrodynamic diameter and ZP after 60 min of AMF exposure.

Due to the non-adiabatic nature of the experimental setup, the SAR was initially determined using the initial-slope method within a restricted time window to ensure quasi-adiabatic conditions, as presented in Figure 5. To quantify heat dissipation, the experimental heating curves for HA@Co-MNP and CMC@Co-MNP ferrofluids were fitted using the Box-Lucas model (Figure S1 and Table S1), yielding estimates of the linear loss parameter. For both samples, the L values were approximately ~1 × 10⁻⁴ W/K, consistent with non-adiabatic setups operating over a temperature range of 20K [23]. By incorporating L, the SAR values for these samples were recalculated to account for linear heat losses to the surroundings. For HA@Co-MNP and CMC@Co-MNP, the Box-Lucas method yielded SAR_Box-Lucas values of 33.1 ± 2.3 W/g and 24.8 ± 2.5 W/g, respectively. These values represent a 15–20% increase over the initial-slope calculations, and the difference in performance induced by the specific surface functionalization with HA or CMC remains significant.

The observed variations in the magnetothermal response of nanoparticles coated with hyaluronic acid and carboxymethylcellulose, which share very similar core characteristics (i.e., size, composition, crystalline structure, single-phase magnetite, etc.), hydrodynamic radii, zeta potential, and biopolymer coating mass ratios, can be fundamentally interpreted by the distinct coordination environments at the organic–inorganic interface provided either by the CMC or the HA chains [12,13,52]. In the superparamagnetic regime of inorganic cores about 5 nm, heat generation is governed by Néel relaxation, which is highly sensitive to the effective magnetic anisotropy, a parameter that is significantly modulated by surface spin pinning [40,53,54,55]. The presence of surface ligands creates a strong field that stabilizes metallic ions on the surface, an effect that propagates into the nanoparticle interior via magnetic exchange interactions. Regarding HA, the perfectly alternating 1:1 periodicity of meres, featuring both carboxyl and acetamido groups, creates a uniform, dense coordination cage through multidentate chelation. The nitrogen atoms within the acetamido groups of HA play a decisive role, as nitrogen possesses a lower electronegativity compared to the oxygen atoms of COOH/COO⁻. Thus, its lone electron pairs are more readily available for coordination with surface metal ions (Co²⁺/Fe²⁺/Fe³⁺), thereby fostering stronger covalency. Indeed, the N 1s XPS spectra shown in Figure 2A reveal this trend. While the N 1s peak for pure HA appears at a characteristic binding energy associated with the secondary amide group, the peak for HA@Co-MNP tends to higher binding energies as a result of the decrease in negative charge on the nitrogen nucleus (deshielding effect), indicating a coordinated interaction between nitrogen and metallic surface ions. These nitrogen-based donors induce greater crystal-field splitting than oxygen-only ligands. Such a modification of the electronic environment of the surface iron d-orbitals effectively tunes the orbital contribution to magnetic anisotropy and, therefore, reduces surface spin disorder, leading to the enhanced heating behavior observed for the HA ligand compared to the CMC-functionalized counterparts [54]. In contrast, although surface passivation is provided by COOH/COO⁻, the relatively random substitution of these chemical groups throughout CMC (DS 0.7) results in a less ordered chemical environment, which could have led to lower effective pinning of surface moments. Although intriguing, a more in-depth investigation of this trend is not the major focus of this study, which would certainly require a sophisticated design and the production of additional nanosystems to address this complex topic.

3.4. Characterization of Catalytic Behavior of Ferrofluids

XPS analysis of the surface of the inorganic core (polysaccharide shell removed by Ar⁺ etching) was performed for the identification of the presence and oxidation states of Fe and Co. It is recognized as a highly “phase-sensitive” characterization technique, which was used to demonstrate the formation of magnetite. The XPS results are depicted in the high-resolution spectra of the Fe region (Figure 6A), which display binding energy peaks at 710.5 eV and 723.7 eV, assigned to Fe 2p3/2 and Fe 2p1/2 peaks of magnetite nanoparticles, shifted by about 0.5 eV towards lower binding energies in comparison to their values in maghemite and hematite [56]. In addition, another piece of evidence supporting the existence of both Fe²⁺ and Fe³⁺ species in the lattice is the absence of a satellite peak neighboring 719 eV, which is clearly observed in maghemite and hematite [56,57,58]. Co²⁺ species were also verified based on the narrow spectra of the Co 2p region (Figure 6B) [59].

The occurrence of these species at the ferrophase surface is usually associated with the peroxidase-like activity of the inorganic core based on Fenton-like reactions (Equations (6) and (7)), schematically depicted in Figure 6C [60,61].

Co-MNP-Metal²⁺_(surface) + H₂O_2(ads) → Co-MNP-Metal³⁺_(surface) + •OH + OH⁻ (ROS)

(6)

Co-MNP-Metal³⁺_(surface) + H₂O_2(ads) → Co-MNP-Metal²⁺_(surface) + •OOH + H⁺ (ROS)

(7)

To confirm the peroxidase-like ability of the ferrofluids, the progress of the blue absorbance of oxidized TMB (TMBox) with time (Figure 6D) upon the injection of hydrogen peroxide substrate (H₂O₂) into ferrofluid suspensions at pH 5.0 ± 0.2 was investigated [62]. This acidic pH range was selected based on the well-known characteristic of the tumor microenvironment (TME) and of endo-lysosomal compartments of cells [6,63,64]. Similar to natural enzymes, a positive dependence of catalytic activity on nanocatalyst concentration (Figure 6E and Figure S2) and substrate amount (Figure 6F and Figure S3) was observed. In addition, they followed the Michaelis–Menten model, characteristic of nanozyme behavior, yielding results that were relatively similar for the Michaelis constant (K_m) and maximum velocity (V_max) (Figure 6G).

For a more in-depth analysis, electron spin resonance (ESR) was employed to monitor the formation of •OH using DMPO as a trapping agent. As illustrated in Figure 6H,I, the characteristic DMPO/•OH spin adduct was observed in the EPR spectra. These results indicated that both HA@Co-MNP and CMC@Co-MNP were capable of generating hydroxyl radicals in the presence of H₂O₂, confirming their peroxidase-like activity [62].

These results demonstrated the potential of ferrofluids for ROS-based catalytic therapy, leveraging their nanozyme-bearing peroxidase-like activity, which was largely independent of the surfactant polysaccharide type.

3.5. Magnetocatalytic Therapy in 2D Cell Model

For 2D cell model experiments, two cell lines were selected to demonstrate the potential of magnetocatalytic therapy to kill GBM cells while preserving non-tumoral cells and to assess the targeting effects of ferrofluids, depending on the surfactant type.

The U-87 MG cell line (likely glioblastoma) was chosen to model human GBM due to its high expression of the CD44 membrane receptor (CD44+). In contrast, HEK 293T (human embryonic kidney cells) is a non-tumoral cell line used as a reference for CD44- [65,66]. According to data extracted from the international source, freely available from the Human Protein Atlas (HPA) project [67,68], the expression level of the CD44 gene in U-87 MG (828.3 nTPM) is approximately 140 times higher than in HEK 293T (5.8 nTPM).

To assess the intracellular levels of ROS, that is, the biocatalytic activity of the ferrofluids, 5-(and-6)-carboxy-2′,7′-dichlorodihydrofluorescein diacetate (DCF-DA) analysis was performed [69,70]. The results (Figure 7A) demonstrated higher ROS levels in the U-87 MG tumoral cell line, although no significant dependence on surfactant type was observed. In addition, it is noteworthy that the lower content of ROS is observed in the absence of the ferrofluids.

The thiobarbituric acid method (TBAR test [71]) was also used to determine malondialdehyde (MDA) formation, an indicator of lipid damage and cell death by ferroptosis, and to assess the biocatalytic activity of ferrofluids. The results (Figure 7B) revealed dependence on both cell type and surfactant, with the highest MDA levels observed in the U-87 MG, as the CD44+ cell line, treated with HA@Co-MNP.

Regarding cell viability associated with the biocatalytic-based therapy (Figure 7C), MTT bioassays showed higher cell viability in the HEK 293T (CD44−) cell line. They depended on concentration, but not on the surfactant polysaccharide (HA or CMC). In U-87 MG (CD44+), a trend toward greater cell killing with the HA stabilizer compared to CMC was observed. These results are summarized in the EC₅₀ (half-maximal effective concentration) analysis, which demonstrates the therapeutic potency (Figure 7D). The lowest EC₅₀ value (0.42 µg/mL) was obtained for the U-87 MG cell line treated with HA@Co-MNP, demonstrating the targeting effect of HA towards cells that overexpress the CD44 receptor. In addition, the selectivity index (SI, ratio between the EC₅₀ of non-tumoral cells and the EC50 of cancer cells) for ferrofluids stabilized by both CMC (SI = 2.0) and HA (SI = 2.4) surfactants indicated a higher “protective effect” to the CD44-deficient non-tumoral cell line [72]. This cytotoxicity of the synthesized anionic-ferrofluids arises from the POD-like catalytic activity of cobalt-doped magnetite core that generates highly toxic hydroxyl radicals (•OH) in the presence of intracellular hydrogen peroxide. Their cytotoxicity underlies biocatalytic therapy, which relies on the overproduction of “exogenous” ROS at very high levels, above the toxic threshold, to eliminate tumor cells (often termed as chemodynamic therapy, CDT) [6,73].

Next, the cancer therapy combining ROS-based and magnetic hyperthermia was studied using the HA@Co-MNP ferrofluid (Figure 7E). As expected, HA@Co-MNP at a concentration of 0.6 µg/mL killed about 32% of U-87 MG tumor cells after 24 h of incubation (ROS+/AMF−). After these 24 h, exposure to an AMF increased cell death to 55% (ROS+/AMF+), representing a 32% decrease in cell viability associated with intracellular heat effects from magnetic-to-thermal conversion. It was noteworthy that the selected concentration of ferrophase (0.6 μg/mL), without AMF (AMF−), was not cytotoxic for HEK 293T non-tumoral cells (targeting effect). Conversely, upon exposure to an AMF, cell viability decreased below the considered cytotoxic threshold (<70%), confirming the significant therapeutic gain of magnetic hyperthermia, a non-invasive focal anticancer strategy. Similar findings were observed in CMC@Co-MNP (Figure 7F).

However, no statistical difference in cell viability was observed between the HEK 293T and U-87 MG cell lines without AMF (ROS+/AMF−, no targeting effect). On the contrary, a lower impact on reducing the cell viability responses to AMF exposure was observed (i.e., 18–22% for CMC@Co-MNP compared with 30–32% for HA@Co-MNP). This effect is likely to be associated with the inferior ferrofluid cell uptake and, consequently, a lower magnetic-to-heat conversion capacity by intracellular CMC@Co-MNP nanohybrids.

3.6. Magnetocatalytic Therapy in 3D Neurospheroids

To closely resemble in vivo solid tumor suppression [64,74], 3D neurospheroids were prepared and treated with two doses of ferrofluid nanotherapeutic at two concentrations (0.6 and 6.0 µg/mL) before exposure to an alternating magnetic field (Figure 8A).

The results of ROS-based therapy for ferrofluids with CMC and HA as surfactants and functional ligands are displayed in Figure 8B. Similar to the 2D cell model, the reduction in cell viability promoted by the POD-like behavior of the ferrofluids indicated a significant difference arising from the surfactant, consistent with tumor targeting via interaction of CD44 membrane receptors with HA at the highest concentration after two doses.

The magnetocatalytic therapeutic effects after the two doses of HA@Co-MNP, followed by exposure to an AMF, clearly demonstrated a reduction in tumor diameter (Figure 8C). Also, cell viability further decreases 24% (0.6 μg/mL) and 51% (6.0 μg/mL), depending on nanozyme concentration, upon magneto-thermal therapy (Figure 8D).

Thus, the results of magnetocatalytic therapy based on 2D cell models and 3D neurospheroids are consistent with the use of combined or multimodal therapies to circumvent the limitations of monotherapy, through additive or synergistic effects [6,75,76,77,78,79]. For systems based on nanozymes (i.e., “enzyme-like nanomaterials”) combined with thermal-induced therapies, some authors have reported a synergistic effect: enhanced nanozyme activity with increasing temperature (i.e., biomimicking enzymes) exhibits catalytic bioactivity that depends on temperature. Also, the synergistic effect of nanozymes at higher temperatures in cancer treatment may arise from accelerated reaction kinetics and diffusion, improved permeability of the tumor microenvironment, enhanced hydrogen peroxide decomposition, reduced glutathione levels, etc. [6,75,76,77,78,79]. In this scenario, the results from magnetocatalytic-driven multimodal anticancer therapy based on Co-doped magnetite ferrofluids showed that the combined effect of both therapies was clearly demonstrated and should therefore be highlighted, with a possible enhancement of ROS generation upon AMF exposure at the intracellular level. Still, current research is not focused on designing and demonstrating synergistic effects (i.e., combined treatments yielding an antitumor effect greater than the sum of their individual effects), but rather on their multimodal anticancer effects.

Thus, these results are highly meaningful, confirming that the HA@Co-MNP ferrofluid offers the opportunity for targeted magnetocatalytic-driven anticancer therapy to enhance the elimination of glioblastoma cells. Figure 9 summarizes the HA@Co-MNP strategy for fighting against GBM tumors.

4. Conclusions

This study reports the successful green synthesis of aqueous ferrofluids based on cobalt-doped magnetite nanoparticles via a co-precipitation strategy, effectively stabilized by two different anionic polysaccharides—carboxymethylcellulose (CMC@Co-MNP) and hyaluronic acid (HA@Co-MNP). Polysaccharides served as both nucleation templates and colloidal stabilizers, rendering a negatively charged supramolecular structure. Comprehensive physicochemical, morphological, and magnetic characterization confirmed the formation of superparamagnetic nanocrystals (mean size of 5–6 nm) with a spinel structure. The magnetic-to-thermal conversion capacity upon exposure to an alternating magnetic field depended on polysaccharide type, with HA-based ferrofluids enabling a higher temperature increase and greater heat dissipation, as evaluated by SAR values. Beyond their hyperthermic capabilities, these ferrofluids behaved as POD-like nanozymes, triggering the generation of cytotoxic hydroxyl radicals from hydrogen peroxide under in chimico and in vitro conditions. In the chemodynamic therapy (CDT) approach, HA-coated ferrofluids demonstrated superior therapeutic efficacy compared to their CMC counterparts, attributed to active targeting of CD44 receptors overexpressed in glioblastoma cells, thereby significantly enhancing cellular lethality in both 2D monolayers and 3D neurospheroid models while sparing non-tumoral cells. The combination of this selective affinity for CD44-overexpressing glioblastoma, bimodal therapy driven by POD-like nanozyme activity under acidic conditions, and externally triggered hyperthermal treatment resulted in enhanced elimination of GBM tumor cells. These scalable, stimuli-responsive ferrofluids open a broad and promising avenue for the future development of precision cancer therapies, aiming to significantly improve patient prognosis and life expectancy.