Theranostic Iron Oxide Nanoparticles for Controlled Oxaliplatin Release Under Simulated Circulation and Cytotoxicity Evaluation in Colorectal Cancer Cell Lines

Abstract

Oxaliplatin (OXA) is a chemotherapeutic agent that suffers from poor pharmacokinetics and off-target toxicity. To enable controlled OXA release, we engineered a multi-functional iron oxide nanoparticle (IONPs) drug delivery system, based on pH-responsive mesoporous Fe₃O₄ (Fe₃O₄@MSN-NH₂) nanoparticles (NPs), conjugated with folic acid (FA) for receptor-mediated targeting and guided by a magnetic robot platform (MRP) under simulated physiologically relevant dynamic circulation/flow system. For FA-conjugated NPs (Fe₃O₄@MSN-NH₂/FA), ~29.73% OXA loading was achieved compared to ~10.3% in controls (Fe₃O₄@MSN-NH₂/OXA), quantified by ICP-OES. Under dynamic circulation flow over 48 h, MRP enhanced pH-responsive OXA release (quantified by HPLC-UV), reaching ~92% and 88% (Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA, respectively) at pH 5, versus 47% and 40% (Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA, respectively) without MRP, demonstrating precise control in acidic tumor-mimicking conditions. MRI relaxometry exhibited strong T2-weighted contrast (T2 = 0.015 s at 50 μg/mL for Fe₃O₄@MSN-NH₂/FA/OXA), confirming theranostic potential. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) studies revealed variable Folate receptor alpha (FOLR1) expression among colorectal cancer cell lines (Caco2, SW620, SW48, and T84), with Caco2 demonstrating high levels. MTT assays indicated selective targeting of FOLR1-positive cells by FA-functionalized NPs (Fe₃O₄@MSN-NH₂/FA). This multi-functional drug delivery system integrates targeted delivery, MRP release, and real-time imaging, offering a promising technique for precision oncology.

1. Introduction

Chemotherapy remains a cornerstone in cancer therapy, with agents such as doxorubicin (DOX), docetaxel, paclitaxel, cisplatin, carboplatin, and oxaliplatin (OXA) improving survival rates [1,2]. OXA is commonly employed in the treatment of gastrointestinal cancers, where it induces immunogenic cell death and is used in both adjuvant and palliative settings [3,4,5,6]. However, its clinical utility is restricted by poor pharmacokinetics and adverse off-target toxicity, including acute neurotoxicity caused by accumulation of platinum (Pt) in spinal ganglion cells, damaging the cell body and axon [7,8,9]. Systemic chemotherapy fails to differentiate between malignant and healthy cells [10], leading to poor prognosis, undesirable side effects, drug resistance, and cancer recurrence [11,12].

To overcome these challenges, targeted drug delivery (TDD) systems have gained attention, aiming to improve drug circulation, increase accumulation in the tumor microenvironment (TME), enhance tissue penetration and cellular uptake, and facilitate controlled drug release (CDR) [13]. Recent developments in nanomedicine have explored tumor-responsive and stimuli-responsive nanoplatforms to improve the therapeutic efficacy and reduce adverse systemic reactions [14,15,16].

Within this evolving field of smart nanomedicines, iron oxide nanoparticles (IONPs) stand out due to their unique magnetic and imaging functionalities. Specifically, magnetite (Fe₃O₄) NPs provide promising features for TDD, because of their biocompatibility, magnetic susceptibility [17,18,19], and their established role as contrast agents for magnetic resonance imaging (MRI) in theranostic platforms that facilitate simultaneous imaging and therapy [20]. They can penetrate anatomical barriers, for instance, the blood–brain barrier, depending on their surface functionalization (e.g., lactoferrin and transferrin ligands) [21]. Drug delivery via IONPs can be achieved via passive targeting depending on NP properties (size, shape, charge and polarity) or active targeting, which involves ligand–receptor interactions [22,23,24,25]. Receptor–ligand interactions are vital in the active targeting of nano-agents [26], involving for instance epidermal growth factor receptors [27], folate receptors (FRs) [28], G protein-coupled receptors [29,30], and fibroblast growth factor receptors [31]. Folate receptor alpha (FOLR1), a glycoprotein with high folate affinity, is highly overexpressed in many carcinomas to support the rapid proliferation of cancer cells [32]. Folic acid, the synthetic form of folate (vitamin B9) [33,34], is widely applied to functionalize IONPs for receptor-mediated uptake and CDR [35,36,37]. In colorectal cancer, FOLR1 expression differs across cell lines such as Caco2, SW620, T84, and SW48, making them appropriate models for studying receptor-mediated targeting [38].

Uncoated Fe₃O₄ NPs face several limitations including aggregation due to high surface energy, susceptibility to oxidation which compromises dispersity and magnetization, and a high chemical reactivity that leads to undesirable interactions [39,40,41]. These drawbacks can be mitigated by surface coating to optimize therapeutics performance both in vitro and in vivo [17,42,43]. Silica (SiO₂) coating can improve dispersibility and enhance drug loading [44,45,46,47], while functionalization with amine groups (-NH₂) facilitates drug attachment and improves circulation time [48]. However, Fe₃O₄ NPs functionalized with SiO₂-NH₂ groups rely on passive targeting for delivery [49,50,51,52,53,54]. Feng et al. [51] demonstrated that Fe₃O₄ NPs functionalized with mesoporous silica NPs (MSNs) and NH₂ groups (Fe₃O₄@MSN-NH₂) and modified with hyaluronic acid enabled active DOX targeting to 4T1 breast cancer cells, improving therapeutic efficacy [51]. In addition to conventional chemical surface functionalization strategies, emerging eco-friendly (green) synthesis pathways provide an alternative method to functionalize NPs directly [55].

Magnetic robot platforms (MRPs) enable wireless navigation of ferromagnetic bodies, including Fe₃O₄ NPs, with high spatial precision and have received significant attention for TDD and CDR in dynamic biomedical environments [56,57,58]. A common configuration involves an electro-permanent magnetic (EPM) platform, which uses an industrial robotic arm equipped with a permanent magnet end-effector. Adjusting the robot’s joints and end-effector position allows optimal EPM positioning for targeted actuation [59]. Although this approach is a promising technique, the integration of MRP for NP navigation and pH-responsive drug release under a simulated in vivo dynamic circulation system remains underexplored. For this purpose, the magnetic responsiveness of NPs is a critical property for MRP applications; however, maintaining it after surface coating and drug loading remains a major challenge [53,54].

Recent studies have investigated Fe₃O₄@MSN-NH₂-based nanocarriers for pH-responsive drug release under static conditions. Notably, Tabasi et al. [53] developed pH-responsive Fe₃O₄@MSN-NH₂/OXA NPs with enhanced OXA loading [53]. The results were promising for the cancer cell lines; however, the release study was conducted under static conditions, which do not adequately represent the dynamic forces that govern NP stability and diffusion behavior in vivo. To the best of our knowledge, no studies have investigated the MRP-controlled release profile of OXA from FA-conjugated Fe₃O₄ NPs under a simulated dynamic circulation/flow system.

This study addresses key shortcomings by utilizing Fe₃O₄@MSN-NH₂-based nanocarriers, conjugating FA targeting ligands and implementing MRP-assisted drug release under continuous circulation. Furthermore, our formulations address issues related to temperature-dependent stability (storage, room, human body, and hyperthermia temperatures), aggregation during functionalization, OXA loading, and release feasibility under MRP actuation. Comprehensive analytical methods were conducted to substantiate functionalization, quantify OXA loading, and validate performance. Gene expression analysis using reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed to quantify relative folate receptor alpha (FOLR1) levels across different cell lines, providing insights into potential targeting strategies. Additionally, the MTT assay was conducted to investigate cell viability following exposure to nanoparticle formulations and oxaliplatin, enabling a comparative assessment of therapeutic efficacy and biocompatibility. Our study offers a substantial step forward compared to previously reported TDD systems.

2. Materials and Methods

2.1. Chemicals and Materials

Anhydrous iron (III) chloride (FeCl₃, AR grade), iron (II) sulfate heptahydrate (FeSO₄·7H₂O, AR grade), cetyltrimethylammonium bromide (CTAB, ≥96.0%), tetraethyl orthosilicate (TEOS, 98%), 3-aminopropyltriethoxysilane (APTES), N-hydroxysuccinimide (NHS, 98%), ammonium nitrate (NH₄NO₃), phosphate-buffered solution (PBS), dimethyl sulfoxide (DMSO), and cellulose dialysis tubing (D9722, 10.0 mm flat width) were obtained from Sigma Aldrich (Dorset, UK). 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDCI), N-Methyl-2-pyrrolidone (NMP, 99%), acetate buffer (AcB), and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (M6494) were acquired from Thermo Fisher Scientific (Birmingham, UK). Ethanol (EtOH) absolute (AR grade, ≥99.8%) and high-performance liquid chromatography grade water (HPLCW) were supplied by VWR International Chemicals (Lutterworth, UK). Ammonium hydroxide (NH₄OH 25–30 wt.% in water, AR grade) was purchased from Merck Life Science UK Limited (Dorset, UK)and OXA was obtained from Molekula Group (Darlington, UK). Folic acid (FA) was purchased from Cayman Chemical Company. RPMI-1640 medium was supplied by Scientistic Laboratory Supplies (Nottingham, UK). Fetal calf serum (FCS) and trypsin were obtained from Fisher Scientific Ltd. (Loughborough, UK) Caco2, SW620, T84, and SW48 human colorectal cancer cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and authenticated by short tandem repeat (STR) profiling prior to use.

2.2. Preparation of Mesoporous Fe₃O₄@MSN-NH₂ Nanoparticles

A detailed overview of chemicals, quantities, experimental conditions, and post-reaction rinsing stages used throughout the synthesis of nanoparticles is available in Supplementary Information Section, Table S1.

Fe₃O₄ NPs were prepared via the co-precipitation method as outlined in our earlier study [60]. Briefly, the precursors, FeCl₃ and FeSO₄·7H₂O, were dispersed in HPLCW and EtOH under N₂ flow and stirred for 45 min at 80 °C. The pH was adjusted to 11 by NH₄OH and the reaction was stopped with no aging. The synthesized Fe₃O₄ NPs were kept in a wet paste form (50% w/w) (WET). The synthesis of Fe₃O₄@MSN-NH₂ was adapted from our recent publication [61]. The WET Fe₃O₄ NPs (1.0 g 50% w/w) were suspended in HPLCW (120.0 mL) and EtOH (60.0 mL), and the solution was sonicated in an ice bath for 45 min. NH₄OH (1.2 mL) was added dropwise under an inert environment and stirred for 1 h at room temperature. Following the dropwise addition of CTAB (0.0729 g) into the solution, TEOS (1.3 mL) was introduced gradually in a controlled manner, and the reaction was left to proceed at ambient conditions for 24 h. The product (Fe₃O₄@CTAB-SiO₂) was rinsed with EtOH and HPLCW multiple times and collected through centrifugation (10,000 rpm, 45 min, 25 °C) and a permanent magnet. A total of 0.560 g (50% w/w) of the WET Fe₃O₄@CTAB-SiO₂ was dispersed in EtOH (50.0 mL), followed by controlled addition of APTES (1.2 mL) while stirring (400 rpm) under N₂ flow at room temperature for 24 h. The resulting product (Fe₃O₄@CTAB-SiO₂-NH₂) underwent several washes with EtOH and was separated by a centrifuge and a permanent magnet. The WET Fe₃O₄@CTAB-SiO₂-NH₂ (0.560 g, 50% w/w) was suspended in EtOH (60.0 mL) and sonicated in an ice bath for 45 min. Separately, NH₄NO₃ (0.006 g/mL, 0.240 g) was dissolved in HPLCW (10 mL) (Supplementary Information Equation (S1)), then it was slowly added to the solution and refluxed for 12 h at 60 °C to remove the CTAB. The reflux temperature of 60 °C was chosen to enable complete removal of CTAB without the compromising the magnetic integrity of the Fe₃O₄ core, based on our previous optimization study [61].

The product (Fe₃O₄@MSN-NH₂) was first washed with HPLCW to remove the surface-bound NH₄⁺ ions and the remaining NO₃⁻; then, EtOH was used to rinse away the CTAB (Equation (2)). The NPs were recovered by centrifugation and a permanent magnet.

2.3. Preparation of Fe₃O₄@MSN-NH₂/FA in the Dark

A total of 0.086 g of EDCI and 0.077 g of NHS were dissolved in 23.0 mL NMP, purged with N₂ gas, and stirred for 1 h. Then, 0.050 g of folic acid was gradually added to the solution, and the reaction was conducted in the dark at room temperature for 24 h. In a separate flask, the Fe₃O₄@MSN-NH₂ (0.560 g, 50% w/w) NPs were dispersed in NMP (85.0 mL) and sonicated, then kept stirred under inert environment and room temperature. The activated folic acid was introduced dropwise to the NP solution and stirred under the same conditions for 48 h. The product (Fe₃O₄@MSN-NH₂/FA) was initially rinsed with EtOH followed by HPLCW and separated by centrifuge and the permanent magnet.

2.4. Drug Loading (Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA)

The WET Fe₃O₄@MSN-NH₂ (0.560 g, 50% w/w) or Fe₃O₄@MSN-NH₂/FA (0.560 g, 50% w/w) was re-dispersed in HPLCW (10.0 mL) and sonicated in an ice bath for 45 min. For each sample, OXA (0.005 g/mL) was separately dissolved in 7.24 mL EtOH and sonicated for 5 min at 45 °C under dark conditions. The OXA solution was added dropwise to each NP solution and stirred (300 rpm) in the dark, under N₂ flow at ambient temperature for 48 h (see Figure 1). The products (Fe₃O₄@MSN-NH₂@OXA and Fe₃O₄@MSN-NH₂/FA/OXA) were rinsed with EtOH and HPLCW and recovered by centrifugation and a permanent magnet. The loading capacity of OXA was investigated by ICP-OES instrument.

2.5. External Magnetic Robot-Assisted Localized Release of Oxaliplatin

A closed-loop dynamic perfusion system was developed to simulate in vivo-like circulation and study OXA release from Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA. In this setup, the NP colloidal suspension continuously recirculated from the reservoir through the pump and into two tissue-mimicking chambers (Chamber 1: tumor mimicking, pH 5.0; Chamber 2: physiological condition, pH 7.4) with and without the MRP assistance for site-specific accumulation and release. In the MRP system, as displayed in Figure 2, a peristaltic pump (Fisher Scientific, Waltham, MA, USA) was connected to a PBS (pH 7.4) reservoir containing 10.0 mL (0.5 mg/mL) of the OXA-loaded samples, suspended in 100.0 mL PBS. The reservoir was placed on a magnetic stirrer plate, operating at a constant temperature of 37 °C and stirring rate of 460 rpm. A controlled flow rate of ~4.0 mL s⁻¹ was selected based on the pump’s preset options and physiological values found in vessels supplying cerebral regions [62] to simulate unidirectional flow through tissue-mimicking environments.

Each environment was localized within a single 3D-printed chamber, housing a tubular cellulose membrane with a cross-sectional diameter of 7 mm under the specified flow conditions. Chamber 1 contained 20.0 mL of acetate buffer (pH 5.0) to simulate an acidic TME. Chamber 2 contained 20.0 mL of PBS (pH 7.4) to represent healthy tissue. A robotic arm (IIWA 14, KUKA, Shelby Township, MI, USA) equipped with an EPM end-effector (neodymium–iron–boron, N52 grade, 1.48 T) was placed at approximately 150.0 mm above Chamber 1 to investigate magnetically guided drug diffusion into acidic tumor conditions as shown in Figure 2 and Video S1. In this setup, the MRP magnetically guided and accumulated NPs within the tumor-mimicking acidic chamber, while simultaneously facilitating localized, pH-responsive OXA release by increasing NP retention and diffusion. Samples (1 mL in volume) were extracted from each chamber at 1, 3, 6, 9, 12, 24, 30, and 48 h intervals. The amount of OXA released under MRP guidance was quantified using high-performance liquid chromatography with ultraviolet detection (HPLC-UV, Agilent Technologies, Santa Clara, CA, USA). The navigation of Fe₃O₄@MSN-NH₂/OXA by MRP is also visualized in Video S2. A summary of the information for the magnetic robot platform (MRP) setup and the dynamic circulation/flow system parameters is provided in Table S2.

2.6. Characterizations

The functionalization of Fe₃O₄ nanoparticles was characterized using a Bruker Vertex 80V FT-IR spectrometer (Bruker Optik GmbH, Ettlingen, Germany), equipped with an RT-DLaTGS detector and a Bruker A225/Q Platinum single-bounce ATR accessory (Bruker Optik GmbH, Ettlingen, Germany) with a diamond crystal. Measurements were performed in the 4000–600 cm⁻¹ range, with a resolution of 4 cm⁻¹ and 32 scans using the attenuated total reflection (ATR) technique. The structural and crystalline phases of the nanoparticles were characterized using a Bruker D2 Phaser powder X-ray diffractometer (PXRD) (Bruker AXS GmbH, Karlsruhe, Germany) equipped with Cu-Kα radiation (λ = 1.541 Å) and a LinxEye detector, operating at 40 kV and 40 mA. Scans were performed over an angular range of 5° ≤ 2θ ≤ 80°, with an exposure time of 10 s per step and a total measurement duration of 11 h. Surface morphology and elemental composition of the nanoparticles were examined using an FEI Tecnai TF20 TEM (FEI Company, Hillsboro, OR, USA) equipped with a field emission gun (FEG), a HAADF detector, an Oxford Instruments INCA 350 EDX (Oxford Instruments, Abingdon, UK) system with an 80 mm X-Max SDD detector (Gatan Inc., Pleasanton, CA, USA), and a Gatan Orius SC600A CCD camera (Gatan Inc., Pleasanton, CA, USA). Dynamic light scattering (DLS) was used to determine the hydrodynamic size (DH), and zeta potential (ZP) was measured to assess the particle charge of the nanoparticles (NPs) in the solution using a Zetasizer (Nano ZSP ZE5600, Malvern Panalytical Ltd., Malvern, UK). For the analysis, the samples were dispersed in HPLCW and homogenized using a Fisher ScientificTM FB705 (Fisher Scientific, Waltham, MA, USA) sonic dismembrator ultra-sonic probe at 30% amplitude with a power output of approximately 210, delivering 2000 J energy over 2 min in continuous mode, while keeping the sample in an ice bath to prevent overheating. The samples were then dispensed in a disposable cuvette and analyzed. The stability of nanoparticle dispersions in PBS was assessed using a Dispersion Analyzer (LUMiSizer 6112-29, 12-channel, ADAPTIVE INSTRUMENTS, LUM GmbH, Berlin, Germany) operating at a wavelength of 865 nm, with light factors ranging from 0.25 to 6. Thermal stability and degradation profiles of the NPs were analyzed by thermogravimetric analysis (TGA) and differential scanning calorimetry (TGA/DSC). The analyses were carried out on Mettler Toledo TGA/DSC 3 + instrument (Mettler-Toledo (Schweitz) GmbH, Greifensee Zurich, Switzerland) between 30 and 700 °C at a heating rate of 5 °C min⁻¹ under N₂ gas. Derivative thermogravimetry (DTG) curves were applied to determine the temperature of maximum weight loss for the NP components. The magnetic properties of the dried nanoparticles were analyzed using a superconducting quantum interference device vibrating sample magnetometer (SQUID-VSM, Quantum Design Inc., San Diego, CA, USA) at 300 K under a magnetic field of 7 T. The platinum content in the samples was quantified by inductively coupled plasma optical emission spectrometry (ICP-OES, Thermo Scientific iCAP7400 radial, Waltham, MA, USA) equipped with a Cetac ASX-520 autosampler (Teledyne CETA, Omaha, NE, USA), the Thermo Qtegra software (Version 2.6, Agilent Technologies), a quartz Seaspray nebuliser, and a quartz cyclonic spray chamber. OXA was quantified using HPLC-UV (Agilent Technologies, Santa Clara, CA, USA) on an Agilent 1290 Infinity II system (Agilent, Santa Clara, CA, USA) equipped with a diode array detector (DAD) set at 254 nm. Chromatographic separation was performed on an Agilent InfinityLab Poroshell 120 EC-C18 column (2.1 × 50 mm, 1.9 µm, Agilent Technologies, Santa Clara, CA, USA) maintained at 40 °C. The mobile phase consisted of 0.1% trifluoroacetic acid (TFA) in water (95%) and 0.1% TFA in acetonitrile (5%). Gradient elution was started with 5% water and was sustained for 5 min at a constant flow rate of 0.5 mL/min. The elemental composition and chemical states of the nanoparticles were analyzed using an ultra-high vacuum (UHV) X-ray photoelectron spectroscopy (XPS) system equipped with an SPECS Phoibos 150 (SPECS Surface Nano Analysis GmbH, Berlin, Germany) hemi-spherical analyzer and 1D-DLD detectors. The X-ray source was a monochromated Al Kα line with a photon energy of 1486.7 eV, operating at 400 W and 15 kV. The vacuum is approximately at 1 × 10⁻⁹ mbar. During the measurements, an electron flood gun is applied to charge neutralize NPs. The dried powder NPs were pressed onto the tape attached to an omicron sample plate. Binding energy was calibrated to the adventitious carbon peak at 284.8 eV. The surface area and porosity of the dried nanoparticle powders were determined using a Brunauer–Emmett–Teller (BET) adsorption–desorption analyzer (Nova 800 series, Anton Paar GmbH, Graz, Austria) operated with Anton Paar Kaomi software (Version 2.0). Relaxometry measurements were performed using a Bruker BioSpec 70/20 USR 7 Tesla preclinical MRI scanner (Bruker BioSpin, Ettlingen, Germany), equipped with a 660 mT/m gradient and a four-element volume array coil (Neos BioTec, Pamplona, Spain). T1 relaxation times were measured using an Inversion Recovery (IR) spin echo (SE) sequence, with the following parameters: echo time (TE) = 4.49 ms, repetition time (TR) = 15,000 ms, matrix size = 128 × 128, field of view (FOV) = 35 × 35 mm, axial slice thickness = 1 mm, and 22 inversion times, varied between 0.00687 s and 7.5 s. T2 mapping was carried out using a multi-echo SE sequence with the same parameters, but comprising 64 echoes without inversion pulse and a TR = 5 s. Relaxation maps were generated using in-house-developed software and region-of-interest (ROI) analysis across phantom samples. Phantoms were prepared by dispersing the NPs at concentrations of 50 µg/mL and 5 µg/mL in Milli-Q water and sonicating in an ice bath for 45 min. Each dispersion was mixed in 1:1 ratio with 1% agarose solution (prepared at 70–80 °C) and transferred into 1 mL syringes to solidify. Syringes were placed in a bespoke 3D-printed sample holder and inserted in the radiofrequency (RF) coil of the MRI system.

2.7. FOLR1 Gene Expression Analysis

Human colorectal cancer cell lines (Caco2, SW620, T84 and SW48 cells) were grown to 70% confluence in the RPMI medium, with 10% (v/v) heat-inactivated FCS. Total RNA was extracted (New England Biolabs, T2010S, Ipswich, MA, USA). cDNA was synthesized from 500 ng RNA using the LunaScript RT SuperMix Kit (New England Biolabs, E3010, Ipswich, MA, USA). Both steps were performed according to the manufacturer’s instructions. Luna Universal qPCR Master Mix kit (New England Biolabs, M3003S, Ipswich, MA, USA) was used for Reverse Transcription Quantitative Polymerase Chain Reaction (RT-qPCR), according to the manufacturer’s instructions, using 1 uL cDNA input. Each reaction was carried out with 1 μL of cDNA input using the QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA). The primers used were FOLR1 forward: AGCACCACAAGGAAAAGCCAGG; FOLR1 reverse: GTGCCATCTCTCCACAGTGGTT; GAPDH forward: TCAACGACCACTTTGTCAAGC; GAPDH reverse: CCAGGGTCTTACTCCTTGG. Relative quantification was obtained using the 2 (^−ΔΔCt) method. Expression values were calculated as log₂ fold-change relative to the T84 cells as the reference for ^ΔΔCt normalization.

2.8. MTT Cytotoxicity Assay

Caco2 and SW48 cells were seeded into 96-well plates at a density of 1000 cells per well in an RPMI medium supplemented with 10% (v/v) heat-inactivated FCS and incubated at 37 °C in a humidified atmosphere containing 5% CO₂. After 24 h incubation, NP formulations and pure OXA were dispersed in culture medium (0.5 mg/mL). The NPs were sonicated for 45 min in an ice bath, while pure OXA was sonicated for 15 min at 45 °C. Serial dilutions of either pure OXA or NP suspensions were then prepared in culture medium. The cells were exposed to either a carrier control or drug/particle concentrations for 72 h. MTT solution (5 mg/mL in PBS) was then added (20 µL/well) to each well and incubated for 4 h. The media were removed, and 150 µL of DMSO was added to dissolve formazan crystals. Once the crystals fully were fully dissolved, absorbance was measured at 620 nm using Cytation^TM 5 Imaging Multimode Reader (BioTek, Winoosi, VT, USA). Statistical analysis was conducted in Prism 9, using the two-way ANOVA test with Geisser–Greenhouse correction. IC₅₀ values were calculated using non-linear regression analysis (log(inhibitor) vs. normalized response, variable slope) in GraphPad Prism (Version 10.0.0, GraphPad Software, Boston, MA, USA), with cell growth normalized to the carrier control.

3. Results and Discussion

Fe₃O₄ NPs were synthesized into mesoporous carriers with amine groups and FA conjugation, facilitating OXA loading. As reported in studies, the application of the functionalized Fe₃O₄ NPs using CTAB surfactant [53,54] paved the pathway for the preparation of coated NPs for the application in TDD and CDR by MRP. However, using CTAB at high quantities (≥0.01 M) and at high temperature (~80 °C) showed reduction in crystallinity and saturation magnetization. The removal of CTAB by calcination [54] was found to cause high aggregation and increase the instability index value. Herein, our work attempts to further improve the reported protocol following optimized steps, including keeping the surfactant until the addition of NH₂ groups. This led to reduced aggregation, as well as improved thermal and dispersion stability of the NPs. Accordingly, the CTAB surfactant was removed via the reflux of Fe₃O₄@CTAB-SiO₂-NH₂ NPs which completely degraded the CTAB, leading to a mesoporous structure. This aligns consistently with the reported literature [51]. As such, our successful procedure for removing the CTAB could be beneficial for increasing the surface area of the functionalized NPs which can ease the process for OXA loading.

3.1. Physicochemical Characterization of Nanoparticles

3.1.1. FT-IR Spectroscopy

FT-IR was applied to validate the stepwise surface functionalization of Fe₃O₄ NPs with SiO₂, NH₂, and FA. The spectra of Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, Fe₃O₄@MSN-NH₂, and Fe₃O₄@MSN-NH₂/FA are shown in Figure 3a(i–iv), respectively. The spectrum of Fe₃O₄@CTAB-SiO₂ (Figure 3a(i)) displays a broad band at ~3400 cm⁻¹, attributing to O-H stretching from adsorbed water or surface hydroxyl groups. Peaks at 2925 cm⁻¹ and 2854 cm⁻¹ are associated with the C-H stretching vibration from the alkyl chain of CTAB. A band at 1637 cm⁻¹ corresponded to the scissoring/bending vibration of adsorbed water (H-O-H bending). The strong peak at 1039 cm⁻¹ is associated with asymmetric stretching of Si-O-Si bonds, while the peak at 1473 cm⁻¹ arises from C-H bending vibrations. The Fe-O vibrations are visible at 630 cm⁻¹. Following the addition of APTES (Figure 3a(ii) Fe₃O₄@CTAB-SiO₂-NH₂), the particles maintained their C-H stretching bands at 2925 cm⁻¹ and 2854 cm⁻¹. The appearance of new signals at 1554 cm⁻¹ and 1546 cm⁻¹ is assigned to N-H bending and confirms the successful addition of amine groups. The Si-O-Si stretch remains visible at 1039 cm⁻¹. In the spectrum of Fe₃O₄@MSN-NH₂ (Figure 3a(iii), the absence of C-H stretching bands at 2925 cm⁻¹ and 2854 cm⁻¹ confirms successful surfactant (CTAB) removal. Upon FA conjugation (Figure 3a(iv), Fe₃O₄@MSN-NH₂/FA), new peaks appear at 3558 cm⁻¹, 3487 cm⁻¹, 3407 cm⁻¹, and 3224 cm⁻¹. Peaks at 3558 cm⁻¹ and 3487 cm⁻¹ are a result of free primary amine (NH₂) stretching vibrations of the pteridine ring of FA. The peak at 3407 cm⁻¹ contributes to N-H stretching of the pteridine ring in FA involved in hydrogen bonding. The peak at 3224 cm⁻¹ suggests the presence of secondary amide N-H stretching, confirming the successful formation of covalent amide bond between the surface MSN-NH₂ and the FA. A prominent peak at 1616 cm⁻¹ corresponds to primary amide C=O stretching vibration which overlapped with the characteristic primary amine bending of the pteridine ring in FA. The Si-O-Si peak at 1060 cm⁻¹ becomes weaker due to surface shielding and chemical modifications introduced by FA. The Fe-O band at 630 cm⁻¹ becomes sharper following FA conjugation, likely due to improved structural organization and decreased interference from surface-adsorbed species.

Table 1. FT-IR spectral analysis of Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, Fe₃O₄@MSN-NH₂, and Fe3O4@MSN-NH2/FA.

Samples	Functional Group	Wavenumber (cm−1)	Assignment	Ref.
Fe3O4@MSN-NH2/FA	N-H	3558	Primary amine stretching of FA pteridine ring	[63,64,65]
	N-H	3487
	N-H	3407	N-H stretching of FA pteridine ring (H-bonding)
	N-H	3224	Secondary amide N-H stretching
Fe3O4@CTAB-SiO2, Fe3O4@CTAB-SiO2-NH2	C-H	2925, 2854	Asymmetric and symmetric C-H stretching of CTAB alkyl chains	[65,66]
Fe3O4@CTAB-SiO2	H-O-H	1637	Scissoring/bending vibration of adsorbed water	[67]
Fe3O4@MSN-NH2/FA	C=O	1616	Primary amide band (C=O stretching) overlapping with FA primary amine bending	[65]
Fe3O4@CTAB-SiO2-NH2, Fe3O4@MSN-NH2	N-H	1554–1527	N-H bending of surface primary amines	[61,63,65,67,68]
Fe3O4@CTAB-SiO2	C-H	1473	Alkyl group bending from CTAB	[63,66]
All samples	Si-O-Si	1083–1039	Asymmetric stretching	[63,69]
All samples	Si-O	800	Symmetric stretching	[61,69]
All samples	Fe-O	630	Stretching vibration of the octahedral Fe-O in Fe3O4	[60]

summarizes the frequency assignments and wavenumbers reported in this study.

The progressive evolution of FT-IR spectra validates the successful synthesis of multi-functionalized Fe₃O₄ NPs. The persistence of Fe-O vibration at 630 cm⁻¹ and the Si-O-Si bands across all synthesis steps (Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, Fe₃O₄@MSN-NH₂, and Fe₃O₄@MSN-NH₂/FA) suggest that the core–shell inorganic structure remains intact despite multiple chemical modifications. The transition from a surfactant (CTAB)-filled SiO₂ shell to a mesoporous structure is evidenced by the disappearance of alkyl C-H stretching bands at 2925 and 2854 cm⁻¹. This is the most critical stage not only for facilitating drug loading but also for reducing cell cytotoxicity associated with residual CTAB. The appearance of N-H bending signals (1554 and 1546 cm⁻¹) confirms successful surface functionalization with primary amines by silanization, facilitating the necessary anchors for ligand attachment. Subsequent FA conjugation is confirmed by the appearance of the secondary amide N-H stretch at 3224 cm⁻¹ and the primary amide band at 1616 cm⁻¹, indicating covalent FA attachment. Such robust covalent linkages guarantee stability during systemic circulation. Furthermore, the presence of the pteridine ring fingerprints (3558 cm⁻¹, 3487 cm⁻¹, 3407 cm⁻¹) indicates that FA retains its structural integrity during coupling process. The observed FT-IR wavenumbers align well with those reported in the literature [60,63,65,66,67,68], validating the successful modification of Fe₃O₄ NPs.

3.1.2. Powder X-Ray Diffraction (PXRD)

PXRD spectra of Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, Fe₃O₄@MSN-NH₂, and Fe₃O₄@MSN-NH₂/FA are shown in Figure 3b, respectively. All samples displayed diffraction peaks at the (111), (220), (311), (400), (422), (511), (440), and (533) crystal planes. The peaks correspond to the diffraction pattern of Fe₃O₄, which is in excellent agreement with ICDD PDF card no. 01-075-0449 and the literature [60].

PXRD patterns of Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, Fe₃O₄@MSN-NH₂, and Fe₃O₄@MSN-NH₂/FA displayed similar diffraction peaks, corresponding to F₃O₄. This indicates that the spinel crystal structure of the core was preserved throughout coating, functionalization, and FA conjugation with no evidence of lattice alteration or secondary phase formation. Similarly, Ni et al. [65] observed that Fe₃O₄@MSN-NH₂/FA maintained their diffraction peaks during FA conjugation and even after paclitaxel loading, confirming structural robustness as this is critical for retaining magnetic characteristics and drug delivery performance [65].

3.1.3. Thermogravimetric and Derivative Thermogravimetric Analysis

The TGA and DTG curves for Fe₃O₄@CTAB-SiO₂ and Fe₃O₄@MSN-NH₂ are displayed in Figure 3c. Both samples showed pronounced weight loss patterns at various temperature regions. As seen, the initial weight loss at a temperature range of ~30–120 °C was attributed to thermal release of the residual water or moisture and other solvents, consistent with the literature [70]. The TGA curves of Fe₃O₄@CTAB-SiO₂ contain four weight loss phases where the second phase of weight loss between ~110 and 300 °C accounted for 19.02% of the total weight loss. This major loss is ascribed to the decomposition of CTAB, similar to the literature [71]. In contrast, a major loss cannot be seen in the case of Fe₃O₄@MSN-NH₂ which supports the removal of CTAB from NPs. Between ~350–700 °C, a smaller weight loss of ~7% was observed, which corresponds to the thermal degradation of APTES molecules. DTG curves supported these findings, exhibiting multiple decomposition stages within the range of 100–400 °C for Fe₃O₄@CTAB-SiO₂ and Fe₃O₄@MSN-NH₂. The detailed thermal data for both samples are summarized in Supplementary Information Table S3.

The thermal studies of Fe₃O₄@CTAB-SiO₂ and Fe₃O₄@MSN-NH₂ by TGA/DTG verifies the synthetic pathway removed CTAB from NPs’ surface. The absence of major weight loss in the 110–300 °C range for Fe₃O₄@MSN-NH₂ supports CTAB decomposition/removal. Jin et al. [72] demonstrated that the residual surfactant can induce cytotoxicity in vitro and in vivo [72]. A minor weight loss was seen between 350 and 700 °C, attributed to the degradation of APTES molecules, confirming successful amine functionalization. These observations align with previously reported studies on mesoporous silica-coated Fe₃O₄ NPs, where removal of CTAB and surface functionalization are vital to obtain colloidal stability and biocompatibility [70,71].

3.1.4. X-Ray Photoelectron Spectroscopy

XPS was used to study the chemical state and electronic environment of Pt species in OXA-loaded samples, Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA, as shown in Figure 4a and Figure 4b, respectively. The Pt 4f spectra for each sample were collected in 10 sweeps with a step size of 0.1 eV, a dwell time of 0.2 s, and a pass energy of 30 eV. Peak fitting was performed using the CasaXPS software (Version 2.3.24, Casa Software Ltd., Teignmouth, UK). with a Gaussian–Lorentzian peak shape (70:30) and a Shirley background. The Pt 4f peaks have two peaks for each chemical state resulting from spin–orbit coupling. This splitting has a distinctive intensity ratio (4:3) and energy separation of approximately 3.3 eV. In the Fe₃O₄@MSN-NH₂/OXA sample, deconvolution of the Pt 4f region (Figure 4a) shows two distinct oxidation states. The Pt 4f_7/2 at 73.59 eV corresponds to partially reduced Pt species (Pt²⁺ or Pt⁰), and Pt 4f_7/2 peak at 77.57 eV can be assigned to oxidized Pt⁴⁺ species. Signal-to-noise ratio was relatively high due to low Pt loading, which limited the significance of this interpretation. Whereas in Fe₃O₄@MSN-NH₂/FA/OXA sample (Figure 4b), a single chemical environment was observed, with the corresponding Pt 4f_7/2 and Pt 4f_5/2 peaks arising from the same oxidation state.

The XPS results showed significant variations in the chemical environment of platinum between Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA formulations. In Fe₃O₄@MSN-NH₂/OXA (Figure 4a), the presence of two Pt oxidation states reveals heterogenous Pt coordination on the amine-functionalized surface. The partially reduced Pt species (Pt²⁺ or Pt⁰) are likely linked to interaction between electron-donating groups on the surface (e.g., amines) and OXA which may have facilitated partial reduction or ligand exchange [73]. The higher binding energy component (Pt 4f_7/2 peak at 77.57 eV) is attributed to oxidized Pt⁴⁺ species, which is consistent with previously reported XPS signatures of OXA-derived Pt environments [74]. In contrast, a single chemical environment in the Fe₃O₄@MSN-NH₂/FA/OXA sample reveals a more homogeneous Pt coordination state, with the Pt 4f_7/2 and Pt 4f_5/2 peaks originating from the same oxidation state. The Pt 4f_7/2 at 72.8 eV is characteristic of the platinum hydroxide species, parallel to the literature [75]. This highlights that under the mild experimental conditions, OXA may have partially dechelated, exposing a reactive Pt site to coordinate with electron-donating groups, such as amine groups or hydroxyl groups present in folic acid.

3.1.5. Transmission Electron Microscopy

Figure 5 shows TEM images of the Fe₃O₄@MSN-NH₂/FA/OXA sample. The TEM images (Figure 5a(i–iii)) exhibited quasi-spherical morphology. The magnetic core size was approximately 7 nm and the overall particle diameter expanded to 14 ± 2 nm after the functionalization and OXA loading (Figure 5d). Selected area electron diffraction (SAED) (Figure 5a(iv)) patterns showed a series of distinct diffraction rings corresponding to magnetite lattice planes, validating the polycrystalline nature and crystalline structure of the Fe₃O₄ core. Energy dispersive X-ray spectroscopy (EDX) mapping confirmed silica coating around the Fe₃O₄ core, forming a core–shell structure (Figure 5b(i)). The EDX spectrum further verified the elemental composition of Fe₃O₄@MSN-NH₂/FA/OXA (Figure 5c).

The TEM analysis confirms the formation of a core–shell structure, where SiO₂ coating provides structural stability and reduces Fe₃O₄ aggregation. The increase in diameter from 7 nm (core) to 14 ± 2 nm after FA conjugation and OXA loading indicates effective surface modification and drug incorporation. SAED patterns verify the crystalline integrity of the NPs post-functionalization, which is essential for retaining magnetic responsiveness. EDX mapping exhibits a consistent silica shell and EDX spectrum confirms the incorporation of FA and OXA onto Fe₃O₄@MSN-NH₂ surface.

3.1.6. Brunauer–Emmett–Teller Surface Area and Porosity Analysis

The specific surface areas of Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA were analyzed by BET nitrogen adsorption–desorption isotherms at 77 K, measured against relative pressure (P/P0) at 77 K. Prior to analysis, the samples were degassed by heating to 60 °C at a rate of 5.0 °C/min for 1 h, followed by further heating to 100 °C at the same rate for 15 h. All samples demonstrated somewhat hysteresis loops, suggestive of mesoporous structure with minor micropores and cylindrical-shaped pores (Figure 6). As expected, the removal of CTAB generated mesoporosity in Fe₃O₄@MSN-NH₂ NPs, which exhibited the highest surface area of 419.95 m²/g and the micropore volume of 0.086 cm³/g. Functionalization with FA (Fe₃O₄@MSN-NH₂/FA) significantly reduced the surface area to 40.33 m²/g, likely due to surface coverage and decreased pore accessibility. The surface area of OXA-loaded samples, Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA, was slightly higher than that of FA functionalized (Fe₃O₄@MSN-NH₂/FA) at 189.06 m²/g and 184.66 m²/g, respectively.

The BET analysis demonstrates that the removal of the CTAB surfactant is vital to generating mesoporosity, as highlighted by the high surface area of Fe₃O₄@MSN-NH₂ (419.95 m²/g). The significant reduction in surface area following FA conjugation (40.33 m²/g) suggests that bulky FA molecules block pore openings and mask available surface sites, which is consistent with a ligand-functionalized mesoporous silica network [76]. Notably, OXA loading partially recovers surface area in both Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA formulations (189.06 m²/g and 184.66 m²/g), which may suggest that OXA molecules adhere to the external surfaces of the NPs instead of penetrating internal mesopores. The findings are fundamental for understanding OXA loading efficiency and release profile.

3.1.7. Dispersion Stability and Colloidal Properties

The dispersion stability of Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/FA, Fe₃O₄@MSN-NH₂/OXA, and Fe₃O₄@MSN-NH₂/FA/OXA in PBS was measured by the LUMiSizer technique at different temperatures: 4 °C (storage), 25 °C (ambient), 37 °C (physiological), and 43 °C (hyperthermia conditions). The instability index, a dimensionless value between zero (highly stable) and one (highly unstable), highlights the tendency of NPs to aggregate/sediment under centrifugal force, providing insight into colloidal behavior under varied conditions [61]. The results are summarized in

Table 2. Dispersion stability of Fe₃O₄@MSN-NH2, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA and Fe₃O₄@MSN-NH₂/FA/OXA in PBS at 4 °C, 25 °C, 37 °C, and 43 °C by LUMiSizer technique.

Sample Name	Instability Index (4 °C)	Instability Index (25 °C)	Instability Index (36 °C)	Instability Index (43 °C)
Fe3O4@MSN-NH2	0.94 ± 0.07	0.79 ± 0.06	0.88 ± 0.05	0.57 ± 0.05
Fe3O4@MSN-NH2/OXA	0.82 ± 0.05	0.90 ± 0.03	0.99 ± 0.06	0.91 ± 0.03
Fe3O4@MSN-NH2/FA	0.83 ± 0.07	1.03 ± 0.09	0.87 ± 0.06	0.76 ± 0.07
Fe3O4@MSN-NH2/FA/OXA	0.80 ± 0.06	0.81 ± 0.05	0.88 ± 0.07	0.51 ± 0.07

and dispersion profiles at 25 °C are shown in Figure 7a. Among the formulations, Fe₃O₄@MSN-NH₂/FA displayed an instability index, exceeding one at 25 °C, while Fe₃O₄@MSN-NH₂/FA/OXA exhibited improved stability across all tested temperatures, and, in particular, at 43 °C, where it outperformed Fe₃O₄@MSN-NH₂/OXA.

Based on LUMiSizer results, FA conjugation independently impacts colloidal stability, likely due to its solubility behavior and shielding of surface charges, which weakens the electrostatic repulsion between NPs [61]. In contrast, the addition of OXA considerably enhances stability, especially under hyperthermic conditions (43 °C). This stabilization is likely due to the contribution of OXA in intermolecular interactions, leading to a reduction in NP aggregation. Overall, the conjugation of FA and OXA onto Fe₃O₄@MSN-NH₂ led to better dispersion stability, in particular, above room temperature, supporting their suitability for physiological environments and potential application in hyperthermia-based therapy.

3.1.8. Dynamic Light Scattering and Zeta Potential

DLS, zeta potential, and polydispersity index (PDI) measurements were applied to determine the colloidal characteristics of the NPs, including their hydrodynamic size, and surface charge in dispersion systems (Figure 7b,c). The hydrodynamic sizes and PDI (shown in brackets) of the Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA were 69 ± 7 nm (0.116), 60 ± 5 nm (0.101), 80 ± 9 nm (0.119) and 93 ± 15 nm (0.113), respectively.

The zeta potential of Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA were 17.35 mV, −32.62 mV, −18.75 mV, and −51.22 mV, respectively. The hydrodynamic size, surface zeta potential, and PDI data of Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA are summarized in

Table 3. Hydrodynamic size, zeta potential, and PDI of Fe₃O₄@MSN-NH₂, F₃O₄@MSN-NH₂/OXA, Fe₃O4@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA at 25 °C.

Parameters	Fe3O4@MSN-NH2	Fe3O4@MSN-NH2/OXA	Fe3O4@MSN-NH2/FA	Fe3O4@MSN-NH2/FA/OXA
Hydrodynamic size (nm)	69 ± 7	60 ± 5	80 ± 9	93 ± 15
Zeta potential (mV)	+17.45 ± 2	−32.6 ± 1.3	−18.76 ± 1.8	−51.2 ± 1.5
PDI	0.116	0.101	0.119	0.113

.

DLS analysis revealed an increase in hydrodynamic size following FA conjugation and OXA loading. The increase in size may be as a result of the formation of the hydration shell due to the interaction between surface functional groups on Fe₃O₄ NPs with surrounding water molecules, during DLS measurements. The low PDI values (<0.12) observed across all formulations indicate good monodispersity and colloidal stability, which are essential for consistent OXA delivery. The slight increase in hydrodynamic size and PDI for the FA and OXA-functionalized NPs points to successful conjugation and steric hindrance introduced by these moieties.

Zeta potential measurements provide understanding on how surface functionalization impacts electrostatic behavior and colloidal stability. In general, particles with a magnitude of greater than ±30 mV are considered to have good colloidal stability as the repulsive electrostatic forces between particles help to suppress the attractive forces, preventing aggregation and rapid sedimentation [77]. Fe₃O₄@MSN-NH₂ (+17.35 mV) displays moderate stability, while OXA loading noticeably shifts to a negative surface charge (−32.62 mV), improving electrostatic repulsion and minimizing aggregation. FA conjugation alone minimized surface charge (Fe₃O₄@MSN-NH₂/FA −18.76 mV), likely due to FA’s zwitterionic profile (containing acidic (-OH) and the basic (-NH₂) functional groups) and the bulky structure of FA masking ionizable surface groups (e.g., NH₂, Si-OH). This effect correlates with the instability index obtained by LUMiSizer analysis (Section 3.1.7), where the Fe₃O₄@MSN-NH₂/FA sample demonstrated a higher instability index at 25 °C. The Fe₃O₄@MSN-NH₂/FA/OXA sample demonstrated the highest magnetite of zeta potential value at −51.22 mV. This substantial increase from −18.76 mV (observed in Fe₃O₄@MSN-NH₂/FA) was primarily due to the chemical structure of OXA. Upon OXA loading, the negative oxalate ligands in OXA likely contributed extra surface charges, increasing the overall negative potential. This led to stronger electrostatic repulsion between NPs, thereby enhancing colloidal stability and decreasing the likelihood of aggregation. As revealed by XPS analysis (Section 3.1.4), OXA may have partially dechelated, exposing its Pt center to coordinate with electron-donating groups (-NH₂) and hydroxyl groups from FA. This coordination likely modified the surface charge, leading to an increased negative zeta potential. Also, FA and OXA collectively may have introduced steric hindrance and enhanced electrostatic repulsion, which amplified surface charge. The significantly high negative zeta potential (−51.22 mV) suggests stronger electrostatic repulsion between particles, which aligns with the improved colloidal stability observed in dispersion studies by LUMiSizer (Section 3.1.7).

3.2. Magnetic Properties

3.2.1. Magnetization and Superparamagnetic Properties

The magnetic properties of the prepared samples were measured using SQUID-VSM at 300 K. As presented in

Table 4. Magnetic properties of the synthesized NPs at 300 K.

Samples	Fe3O4	Fe3O4@CTAB-SiO2	Fe3O4@CTAB-SiO2-NH2	Fe3O4@MSN-NH2	Fe3O4@MSN-NH2/OXA	Fe3O4@MSN-NH2/FA	Fe3O4@MSN-NH2/FA/OXA
Hc (T)	0.00060 (2)	0.00142 (2)	0.00040 (2)	0.00048 (2)	0.00038 (2)	0.00012 (2)	0.00023 (2)
Mr (emu/g)	0.664	0.548	0.385	0.398	0.316	0.229	0.1098
Ms (emu/g)	78.50	27.52	58.85	61.24	55.35	37.48	38.55
Mr/Ms	0.00846	0.01991	0.00654	0.00650	0.00571	0.00610	0.00495

Remnant magnetization (Mr); coercive field (Hc); saturation magnetization (Ms); remnant magnetization (emu/g)/saturation magnetization (emu/g) (Mr/Ms). For Hc values, uncertainty is shown based on instrument precision, while other parameters (Mr, Ms, Mr/Ms) exhibited negligible uncertainty compared to their magnitude and are displayed as mean values.

, all samples showed negligible coercivity (Hc ~0.00012–0.00142 T) and low remanent magnetization (Mr/Ms ≤ 0.02). The saturation magnetization (Ms) values of Fe₃O₄, Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA NPs were 78.50 emu/g, 27.50 emu/g, 58.84 emu/g, 61.12 emu/g, 55.35 emu/g, 37.48 emu/g, and 38.55 emu/g, respectively. The complete magnetic parameters for all samples are collected in Table 4; Figure 8a,b illustrates saturation magnetization curves for Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA samples at 300 K.

3.2.2. Frequency Dependence of the Blocking Temperature

The blocking temperature (Tb) was obtained from the peak position of the imaginary component of AC susceptibility (ACX″) versus temperature. ACX″ data for Fe₃O₄, Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, and Fe₃O₄@MSN-NH₂ are provided in the Supplementary Information Section (Figure S1), where each sample exhibits a characteristic Tb corresponding to ACX″.

To analyze the magnetic relaxation dynamics and the frequency dependence of the Tb, the natural logarithm of the inverse relaxation time ($\mathrm{l}\mathrm{n}(1/\tau )$) was plotted against Tb (K), as shown in Figure 8c (the mathematical calculations are collected in Supplementary Information Table S4).

The Nèel–Brown equation was used to extract relaxation parameters [78]:

$$\tau ={\tau }_{0}exp({\displaystyle \frac{U}{k_{B}T}})$$

(1)

where

τ = relaxation time

τ₀ = attempt time

U = magnetic anisotropy energy barrier

K_B = Boltzmann constant (1.380 × 10⁻²³ J/K)

T = temperature (K)

Negligible coercivity (field strength needed to reverse magnetization to zero after saturation) and low remanent magnetization (magnetization remaining after removal of external magnetic field) indicate minimal interparticle magnetic exchange and weak dipole–dipole interactions, consistent with the previously reported literature [79]. These characteristics confirm superparamagnetic nature, in agreement with the literature [54]. The absence of magnetic remanence indicates that the NPs do not retain magnetization once the external magnetic field is removed, which is essential for biomedical utilization that relies on magnetic responsiveness without permanent magnetization. The significant reduction in saturation magnetization for Fe₃O₄@CTAB-SiO₂ can be attributed to the presence of CTAB, which acts as a non-magnetic dead layer on the NP surface, disrupting spin alignment and weakening magnetic coupling. This phenomenon is consistent with previous studies [50,80,81]. SiO₂ and NH₂ coatings also introduce isolating effects, which can lead to a reduced coercivity as noted in the literature [70,82]. FA conjugation and OXA loading further decreased saturation magnetization due to surface coverage. Nevertheless, the final saturation values (37.48 emu/g and 38.55 emu/g) remain within the effective range for biomedical applications. According to Nguyen et al. [83], Fe₃O₄ NPs with saturation magnetization values as low as 30–40 emu/g have proven effective in biomedical applications, including magnetic targeting, MRI, and magnetic hyperthermia. The S-shaped magnetization curves without appreciable hysteresis (Figure 8a) demonstrate that thermal energy at 300 K is adequate to overcome the magnetic anisotropy energy barrier, resulting in rapid fluctuations of magnetic moment once the external field is removed. This observation confirms the superparamagnetic properties of the synthesized NPs.

The linear trends across the measured temperature range (Figure 8c) confirm that Fe₃O₄, Fe₃O₄@CTAB-SiO₂, Fe₃O₄@CTAB-SiO₂-NH₂, and Fe₃O₄@MSN-NH₂ samples follow a thermally activated relaxation process, enabling calculation of the magnetic anisotropy energy barrier ($U$) and attempt times (${\tau }_0$), applying Arrhenius-based analysis [79]. In single domain NPs, thermal energy facilitates magnetic moment flips across the anisotropy energy barrier ($U$). Above the Tb, the magnetic relaxation becomes rapid enough that the magnetization averages to zero during measurements, demonstrating superparamagnetic behavior [78,84]. The presence of CTAB decreases magnetic coupling and dipole–dipole interactions by reducing aggregate on and improving colloidal stability. Following the completion of the functionalization and subsequent removal of CTAB, the magnetic behavior, especially saturation magnetization, improved, approaching values observed for uncoated Fe₃O₄ core NPs. The observed frequency-dependent change in Tb indicates the dynamic relaxation results align well with the static magnetization trends (Section 3.2.1), confirming that magnetic characteristics are controlled by thermal activation rather than strong interparticle exchange coupling. Collectively, these findings confirm that all synthesized NPs demonstrate superparamagnetic profile under the tested conditions.

3.3. MRI Analysis

To evaluate the MRI contrast potential of the synthesized Fe₃O₄-based therapeutic agents, both longitudinal (T1) and transverse (T2) relaxation times were measured at 50 µg/mL and 5 µg/mL concentrations. Across all samples, T2-weighted contrast was the predominant effect, while T1 values remained relatively consistent (~2.25–2.85 s), suggesting minimal influence on longitudinal relaxation. The data are provided in

Table 5. Relaxation times (T1 and T2) of the Fe₃O₄@MSN-NH₂, F₃O₄@MSN-NH₂/OXA, Fe₃O4@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA at 5 µg/mL and 50 µg/mL concentrations in 1% agarose phantom.

ROI	Samples	[µg/mL]	T1/[s]	T2/[s]
1	Fe3O4@MSN-NH2/FA	50	2.254± 0.047	0.013107 ± 0.00049
2	Fe3O4@MSN-NH2/OXA	50	2.368 ± 0.050	0.03400 ± 0.00084
3	Fe3O4@MSN-NH2	50	2.615 ± 0.064	0.0688 ± 0.0017
4	Fe3O4@MSN-NH2/FA	5	2.818 ± 0.048	0.05624 ± 0.00067
5	Fe3O4@MSN-NH2/FA/OXA	5	2.835 ± 0.055	0.05848 ± 0.00064
6	Fe3O4@MSN-NH2/OXA	5	2.819 ± 0.045	0.08001 ± 0.00079
7	Fe3O4@MSN-NH2	5	2.852 ± 0.054	0.09014 ± 0.00085
8	Fe3O4@MSN-NH2/FA/OXA	50	2.382 ± 0.054	0.01544 ± 0.00050
9	Agarose	-	2.826 ± 0.051	0.07295 ± 0.00063

.

The T2-map analysis (Figure 9) demonstrated significant contrast enhancement in OXA-loaded and FA-functionalized samples. Notably, at higher concentration (50 µg/mL), Fe₃O₄@MSN-NH₂/FA (ROI 1) showed the shortest T2 at 0.0131 s, followed closely by Fe₃O₄@MSN-NH₂/FA/OXA (ROI 8) at 0.0154 s. These samples were visualized as the darkest regions in the T2 map, reflecting their superior transverse relaxation efficiency. In contrast, unmodified Fe₃O₄@MSN-NH₂ at both concentrations (ROI 3: 0.0688 s at 50 µg/mL; ROI 7: 0.0901 s at 5 µg/mL) exhibited significantly longer T2 values, indicating weaker contrast.

At 5 µg/mL concentrations, Fe₃O₄@MSN-NH₂/FA (ROI 4) and Fe₃O₄@MSN-NH₂/FA/OXA (ROI 5) samples retained moderate contrast, while Fe₃O₄@MSN-NH₂/OXA (ROI 6) and Fe₃O₄@MSN-NH₂ (ROI 7) demonstrated poor performance.

The MRI studies verified that surface functionalization substantially impacts transverse relaxation efficiency. FA and OXA loading synergistically improved T2 contrast, likely resulting from slightly increased NP clustering and enhanced local magnetic inhomogeneities, which are key contributors known to intensify T2 relaxivity [85]. This effect was most apparent at higher concentrations (50 mg/mL), where Fe₃O₄@MSN–NH₂/FA and Fe₃O₄@MSN–NH₂/FA/OXA attained T2 values comparable to current clinical agents. These findings align with clinical iron oxide contrast agents such as Resovist, Ferumoxytol, and Ferumoxtran-10 (T₂ range: 0.008–0.030 s), positioning Fe₃O₄@MSN-NH₂/FA/OXA as a promising candidate for MRI-guided targeted chemotherapy. The strong T₂ contrast, combined with pH-responsive drug release and magnetic navigation, underscores the theranostic potential of this multi-functional platform.

The relatively consistent T1 values (~2.25–2.85 s) across all formulations suggest that longitudinal relaxation largely remained unaffected by surface modifications, while substantiating that these NPs primarily perform as T2 contrast agents.

3.4. Drug Loading Capacity

The concentration of Pt was quantified by ICP-OES to determine the OXA loading capacity in Fe₃O₄@MSN-NH₂/OXA, and Fe₃O₄@MSN-NH₂/FA/OXA using the commercial OXA as reference. To prepare samples for the analysis, the NPs were dissolved in 10 mL aqua regia and heated for 2 h. HPLCW was added to adjust the final volume to 50 mL, from which a 1 mL aliquot was taken for analysis. The results showed the Pt contents of Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA were 5.07% and 14.60%, respectively. The drug loadings of Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA were 10.33% and 29.73%, respectively (considering molecular weight of OXA is 397.29 g/mol). The FA-functionalized (Fe₃O₄@MSN-NH₂/FA/OXA) sample had a three-fold increase in OXA loading efficiency compared to the non-FA-functionalized (Fe₃O₄@MSN-NH₂/OXA) sample.

3.5. Drug Release with Magnetic Robot Actuated/Without Magnetic Robot

3.5.1. Overview of Release Profile and System Setup

The drug release profile was evaluated in previously explained the closed-loop perfusion system (Section 2.5, simulating physiological (pH 7.4) as well as tumor-mimicking (pH 5.0) environments, under controlled dynamic flow (~4.0 mls⁻¹) and MRP for site-specific NP accumulation (Figure 10a,b). The release was quantified by HPLC-UV (OXA was used as reference in HPLC-UV analysis), and the results are displayed in Figure 10c,d. The tubing dimensions (210.0 cm total length of the flow path with 5.0 mm inner diameter of plastic tubing and a 10.0 cm dialysis section with inner diameter of 7.00 mm) reflect engineering constraints, rather than physiological vessel sizes. Physiological relevance [86,87,88] and reproducibility were ensured through flow rate (Q) control at ~4.0 mL·s⁻¹ (4.0 × 10⁻⁶ m³/s), to produce laminar velocities and shear stresses comparable to microvascular ranges. The resulting Reynold number and flow regime were calculated as follows:

$$Re= {\displaystyle \frac{\rho vD}{u}}$$

(2)

where

Re = Reynold number (dimensionless)

$\rho$ = fluid density (kg m⁻³), (PBS: 1000 kg m⁻³ (25–37 °C))

$v$ = average velocity (m s⁻¹)

D = diameter of tubing (m)

u = dynamic viscosity (kg m⁻¹ s⁻¹)

Plastic tubing (D = 0.005 m)	A ≈ 1.96 × 10−5 m2, v = Q/A ≈ 0.204 m s−1, Re = 1020
Dialysis section (D = 0.007 m)	A ≈ 3.85 × 10−5 m2, v = Q/A ≈ 0.104 m s−1, Re = 728

The values of Reynolds number at 1020 (plastic tubing) and 728 (dialysis tubing) fall below the turbulence threshold of Re ≈ 2000 [89]. This confirms a stable laminar flow regime that minimizes shear-driven coating loss.

Although the Raynolds numbers are relatively high and comparative to large artery environments, wall shear stress is a more essential factor to consider in the dynamic flow setup, since it represents the frictional force per unit area imposed by fluid flowing along tubing. This could physically strip OXA molecules and coating from the surface, leading to a faster release rate than the static system. To calculate wall shear stress (t_w) [90], the following formula was used:

$$t_w= {\displaystyle \frac{8uv}{D}}$$

(3)

where

t_w = wall shear stress (Pa, N·m⁻²)

u = dynamic viscosity (Pa·s), (PBS at 25 °C = 0.001 Pa·s and at 37 °C = 0.0007 Pa·s)

v = average velocity (m·s⁻¹)

D = diameter of tubing (m)

The wall shear stress values for plastic tubing were ~ 0.33 Pa (u = 0.001 Pa·s) and 0.23 Pa (u = 0.0007 Pa·s) and the values for dialysis tubing were 0.12 Pa (u = 0.001) and 0.08 Pa (u = 0.0007). These wall shear stress values align within microvascular/low shear conditions (0.1–1.5 Pa), making them representative of microvascular conditions [90].

3.5.2. MRP-Assisted Release Profiles

Under closed-loop dynamic perfusion system with MRP (Figure 10a), NPs gradually accumulated in the acidic chamber where the MRP was positioned, over a 48 h circulation period. Upon removal of the MRP from the acidic chamber (Figure 10b), the accumulated NPs were re-dispersed into the circulation system, indicating preserved superparamagnetic characteristics throughout the experiment. At physiological pH (7.4), both nano-formulations demonstrated OXA release below the detection limit over 48 h (Figure 10c), suggesting high OXA retention and colloidal stability throughout circulation. Under acidic conditions (pH 5.0), both samples exhibited significant pH-responsiveness release. Notably, Fe₃O₄@MSN-NH₂/OXA exhibited a faster initial release, reaching roughly 50% of OXA within 5 h and approximately 92% by 48 h. Conversely, Fe₃O₄@MSN-NH₂/FA/OXA demonstrated a more gradual release profile, reaching about 18% at 5 h and around 88% by the end of 48 h (Figure 10c).

3.5.3. Release Profiles Without MRP

In the absence of the MRP (Figure 10d), under the same dynamic perfusion condition, the release profiles of Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA were controlled solely by pH-release conditions. Under physiological pH (7.4), both samples demonstrated minimal OXA release (<10%) over 48 h. At acidic (pH 5.0) environments, both samples exhibited more substantial OXA release, with approximately 47% for Fe₃O₄@MSN-NH₂/OXA and 40% Fe₃O₄@MSN-NH₂/FA/OXA. These findings verify the pH-responsiveness profiles of both samples and demonstrate the extra advantage of MRP-assisted navigation for controlled OXA release.

3.5.4. Mechanistic Interpretation

The drug release studies in the dynamic circulation system by MRP exhibited that both formulations (Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN–NH₂/FA/OXA) have good stability under physiological environments (pH 7.4; chamber 1), which is important for preventing premature OXA leakage during systemic circulation. The low OXA release at pH 7.4 (<10% without MRP) verifies strong OXA–carrier interactions and stable colloidal profile. Under acidic conditions (AcB, pH 5.0), both samples demonstrated substantial OXA release, confirming their pH-responsive behavior. The faster OXA release observed for Fe₃O₄@MSN-NH₂/OXA is attributed to protonation of surface -NH₂ groups which reduces their interactions with OXA and enhances its diffusion from the mesoporous structure. On the other hand, observed slower release for Fe₃O₄@MSN–NH₂/FA/OXA is likely attributed to steric hindrance and stronger binding interactions between the Pt center of OXA and electron-donating moieties in FA, as supported by XPS data.

3.5.5. Contribution of MRP Navigation

The MRP system exhibited spatially (site-specific accumulation) and temporally (extended, regulated release) controlled OXA release, enabling NPs’ accumulation and sustained OXA diffusion over 48 h. The results showed that the application of MRP markedly improved OXA release under an acidic environment compared to without the MRP system. At pH 5.0, Fe₃O₄@MSN-NH₂/FA/OXA released ~88% under MRP control versus ~40% without MRP during 48 h (Figure 10c,d). This corresponds to more than a twofold enhancement in release efficiency attributable to MRP assistance. The enhancement reflects MRP-assisted NP accumulation in the acidic chamber, which maximized residence time and sustained OXA diffusion over the 48 h circulation period. While in the absence of MRP, OXA release kinetics were slower and had lower cumulative release, highlighting the critical role of MRP navigation in achieving controlled delivery under dynamic circulation conditions. Ultimately, this multi-responsive system offers a major advantage for enhancing drug accumulation at TME, while reducing the off-target exposure, highlighting the potential of these NPs for MRP-navigated cancer therapy.

3.5.6. Comparative Analysis with Literature

Table 6. Comparison of drug loading, release profile, types of release, and targeting modality for Fe₃O₄-based nano-formulations under dynamic circulation and static conditions.

Formulations	Payload	Loading Metric and (Quantification Method)	Release Profile (48 h)	Type of Release Study	Targeting Modality	Ref.
Fe3O4@MSN-NH2/FA/OXA	OXA	DL: 29.73% (ICP-OES)	pH 5.0: 88.0%, pH 7.4: ~0.0% (MRP) pH 5.0: ~40.0%, pH 7.4: <10.0% (no MRP)	Dynamic circulation ± MRP	FA-FOLR1 (active) + MRP	This study
Fe3O4@MSN-NH2/OXA	OXA	DL: 10.33% (ICP-OES)	pH 5.0: ~94.0% (MRP), pH 7.4: ~0.0% (MRP) pH 5.0: ~47.0%, pH 7.4: <10.0% (no MRP)	Dynamic circulation/flow ± MRP	NH2 surface + MRP	This study
Fe3O4@MSN-NH2/TPP/OXA	OXA	DL: 16.29% (ICP-OES)	pH 5.0: ~94.3% pH 7.4: 9.0%	Static-no flow	TPP mitochondrial targeting	[61]
Fe3O4@MSN-NH2/OXA	OXA	DL: 10.33% (ICP-OES)	pH 5.0: ~88.4% pH 7.4: <10.0%	Static-no flow	NH2 surface	[61]
Fe3O4@MSN/OXA	OXA	EE: 18.50% (ICP-OES)	pH 5.0: >93.0% pH 7.4: <10.0%	Static-no flow	Unmodified MSN	[53]
Fe3O4@MSN-NH2/OXA	OXA	EE: 55.00% (ICP-OES)	pH 5: >93.0% pH 7.4: <10.0%	Static-no flow	NH2 surface	[53]

Drug loading (DL); encapsulation efficiency (EE).

compares drug loading, release profile, types of release, and targeting modality across multi-responsive Fe₃O₄ systems and selected pH-responsive Fe₃O₄-based nano-formulations reported in the literature, under dynamic circulation and static conditions. Although conventional static drug release studies, such as Tabasi et al. [53], achieved high release at pH 5.0 (>93%), the system was evaluated under static conditions without magnetic guidance or a dynamic circulation system, reducing its physiological relevance. Similarly, Moeni et al. [61] designed TPP-functionalized the Fe₃O₄@MSN–NH₂/OXA system for mitochondrial targeting, which displayed pH-responsive release of about 94%, but was also evaluated under static conditions, without dynamic perfusion or magnetic assistance. These constraints contrast with our system, which uniquely integrates closed-loop dynamic perfusion setup with MRP navigation and controlled drug release. Unlike conventional static pH-responsive systems, our formulations (Fe₃O₄@MSN–NH₂/OXA and Fe₃O₄@MSN–NH₂/FA/OXA) combine dynamic circulation with magnetically navigated MRP, enabling NP site-specific accumulation and controlled release. This system demonstrated no premature OXA release at simulated physiological pH 7.4 and more-than-twofold-higher release efficiency under tumor-simulated conditions (pH 5.0) compared to dynamic perfusion with no MRP.

3.6. Proof of Principle of Targeted Cytotoxicity In Vitro

3.6.1. FOLR1 Gene Expression in Colorectal Cancer Cell Lines

We tested a panel of human colorectal cancer cell lines (Caco2, SW620, T84 and SW48 cells) for FOLR1 gene expression (FOLR1-positive) using RT-qPCR (Figure 11). All four cell lines demonstrated detectable levels of FOLR1 expression. Among them, Caco2 displayed the highest FOLR1 expression (FOLR1^high or FOLR1-positive), while SW48 cells exhibited the lowest FOLR1 expression (FOLR1^low or FOLR1-negative). Based on these results, Caco2 and SW48 were selected for further NP testing to represent FOLR1-positive and FOLR1-negative.

In gene expression studies, the detected variability in FOLR1 expression among colorectal cancer cell lines (Caco2, SW620, T84 and SW48) validates the rationale for choosing FA-functionalized NPs for targeted OXA delivery. Caco2 cells with FOLR1^high expression provide a suitable model for evaluating receptor-mediated uptake, whereas SW48 cells serve as FOLR1^low expression control. These data were comparable to that of the publicly available database, The Human Protein Atlas [91], which reports high FOLR1 gene expression in Caco2 cells and undetected levels in SW48 cells. The Human Protein Atlas [91] confirms that the protein-level expression of FOLR1 in Caco2 and SW48 cells follows the same trend as gene expression results [57]. These distinct expression patterns establish a foundation for investigating whether FA conjugation improves NPs’ specificity and cytotoxicity in FOLR1-positive cells, which is a critical step toward verifying targeted therapeutic potential.

3.6.2. Cytotoxicity of Functionalized Nanoparticles

We then tested the cytotoxicity of Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA in FOLR1^high Caco2 and FOLR1^low SW48 cell lines. Fe₃O₄@MSN-NH₂ (Figure 12a) treatment on both cell lines showed a similar trend, but toxicity was observed at higher concentrations, due to the Fenton reaction [92,93,94]. However, FA-conjugated (Fe₃O₄@MSN-NH₂/FA) NPs showed not much toxicity with FOLR1^low SW48 cells (Figure 12c, p < 0.0001), while significantly greater cytotoxicity towards FOLR1^high Caco2 cells. To establish baseline OXA sensitivity, both cell lines were exposed to pure OXA (SW48 cells were markedly more sensitive to OXA than Caco2 cells) (p < 0.0001), and results are presented in Supplementary Information Figure S2. In Fe₃O₄@MSN-NH₂/OXA (Figure 12b), both Caco2 and SW48 cells displayed dose-dependent growth, with no distinct selectivity between the two cell lines. Also, Fe₃O₄@MSN-NH₂/FA/OXA (Figure 12d) inhibited proliferation in Caco2 and SW48 cell lines in a dose-responsive manner; however they had similar responses, suggesting that OXA sensitivity outweighed the contribution of FA-mediated uptake. The IC₅₀ values for Fe₃O₄@MSN-NH₂, Fe₃O₄@MSN-NH₂/OXA, Fe₃O₄@MSN-NH₂/FA, and Fe₃O₄@MSN-NH₂/FA/OXA are presented in Table S5.

The high cytotoxicity of FA-conjugated NPs in FOLR1^High Caco2 cells verifies that FA conjugation enhances receptor-mediated targeting (Figure 12c). The data validates the hypothesis that folate receptor expression affects NPs uptake and therapeutic response. However, in Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA samples (Figure 12b and Figure 12d, respectively), Caco2 and SW48 displayed very identical dose-dependent curves; hence, any FA-sensitive selectivity is not evident in OXA-loaded formulations. In combination with the differences in OXA loading (Section 3.4) between Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA, these results highlight the requirement for further studies, particularly on alternative colorectal cancer cells that demonstrate differential FOLR1 gene expression levels but comparable OXA sensitivity.

It should be noted that Fe₃O₄-based drug delivery systems are known to accumulate in reticuloendothelial system (RES) organs and undergo gradual degradation [43], thereby necessitating comprehensive future studies on biodistribution, clearance, and long-term safety prior to clinical advancement.

Importantly, our MRP system is strategically intended to minimize off-target RES accumulation by actively navigating and controlling superparamagnetic nanocarriers at the tumor site. This strategy mitigates the natural tendency of NPs to localize in the liver and spleen, thereby improving their translation safety profile. Future work will focus on the comprehensive evaluation of OXA-loaded samples (Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA) under MRP-assisted navigation, and magnetically induced hyperthermia performance under dynamic circulation platform, which will be reported in a forthcoming publication.

4. Conclusions

This study presents a novel approach for spatially and temporally controlled release of OXA using an MRP. The system employs Fe₃O₄@MSN-NH₂/OXA and Fe₃O₄@MSN-NH₂/FA/OXA nanoparticles. Extensive characterizations (FT-IR, XRD, XPS, LUMiSizer, DLS, zeta potential, VSM, and BET) confirmed successful fabrication, sequential surface functionalization, and mesoporous structure. All samples maintained their superparamagnetic characteristics. Compared to the Fe₃O₄@MSN-NH₂/OXA sample, FA-conjugated NPs (Fe₃O₄@MSN-NH₂/FA/OXA) exhibited excellent colloidal stability, strong T2-weighted contrast, and a three-fold-higher OXA loading efficiency (up to 29.7%), showing the advantages of FA conjugation. Additionally, HPLC-UV analysis confirmed MRP-controlled release with minimal/no premature leakage and improved site-specific delivery, which yields substantially higher drug release compared to a non-MRP-controlled system, ensuring sustained, pH-responsive release.

Building on this proof-of-concept work in a simulated TME, cytotoxicity assays on colorectal cancer cell lines with differential FOLR1 expression exhibited FOLR1-associated selectivity of FA-conjugated NPs, supporting receptor-mediated cellular uptake. To our knowledge, this is the first report to combine FA-conjugated Fe₃O₄@MSN-NH₂ NPs with an MRP for controlled release of OXA under physiologically relevant dynamic circulation/flow systems. This integrated strategy introduces a novel avenue for the design of MRP-navigated, ligand-targeted theranostic nanomedicines.

These initial biological evaluations support the platform’s therapeutic potential and represent a significant advance toward further in vitro and in vivo validation. As the next step, we propose studies in tumor-bearing models under MRP navigation to evaluate therapeutic efficacy and translational potential. In summary, this work lays the foundational framework for the development of magnetically navigated, targeted chemotherapeutic drug delivery systems with potential applications in precision oncology.