Synthesis and Characterization of λ-Carrageenan Oligosaccharide-Based Nanoparticles: Applications in MRI and In Vivo Biodistribution Studies

Abstract

λ-type carrageenan (λ-CAR) polysaccharides remain overlooked in the preparation of medical nanoparticles (NP) due to their unsuitable rheological properties and undesired biological effects, although they can also offer advantageous properties. To overcome these obstacles, oligosaccharide derivatives (λ-COS) have been successfully applied to the synthesis of stable NP incorporating both, ferrite cores and divalent manganese (Mn²⁺) releasable ions. The acute pro-inflammatory behavior and anticoagulant activity of native λ-CAR were significantly reduced in the case of λ-COS and λ-COS NP, rendering possible their use for medical applications. In vivo MRI studies in mice showed that the λ-COS NP framework is promising for two applications. The first is partial Mn²⁺ release into the plasma to achieve intracellular Mn²⁺-based contrast of the myocardium and imaging of the hepatobiliary system. Second, it serves as a novel sugar-based coating that confers suitable pharmacokinetic properties to NP, making it promising for further targeted therapy applications.

1. Introduction

Polysaccharides are widely used as coatings or scaffolds for the design of multifunctional nanoparticles (NP) intended for advanced biomedical applications, such as targeted, controlled drug delivery and diagnostics.¹⁻³ They are mostly biocompatible and can ensure a high colloidal stability to the NP, or build controlled release systems, while being easily functionalized. Furthermore, they can impart NP with additional advantageous features, such as targeting abilities or specific bioactivities.⁴ However, applications often focus on particular polysaccharides varieties such as dextran⁵ or chitosan,⁶ neglecting numerous other families with unexplored potential, such as λ-carrageenan (λ-CAR).

λ-CAR belongs to the broad family of carrageenans (CARs), a group of high-molecular-weight marine sulfated polygalactans (up to 300 kDa) derived from red algae.⁷ They are characterized by long, homogeneous, linear chains of repeated disaccharide units consisting of a 1,3-linked β-d-galactopyranose (G unit), alternating with a 1,4-linked α-d-galactopyranose (D unit). Various sulfated substitutions and the presence of a 3,6 anhydro-bridge on the D unit, shape the different subtypes, the most prominent of which are kappa (κ), iota (ι), and lambda (λ). Carrageenans have been extensively used in the food⁸ and cosmetic industries⁹ as thickeners and texturing agents because of their unique rheological properties. They are also often cited for their specific binding of cations, making them valuable for media decontamination applications.¹⁰⁻¹² Furthermore, Carrageenans have been investigated for valuable bioactivities, including antitumor, antimicrobial, and immunomodulatory effects.^7,13 However, their lack of innocuousness and concomitant presence of undesirable biological properties have strongly limited their pharmacological in vivo applications, especially for systemic administration. Specifically, further clinical applications of λ-CAR can be limited by its viscosity and, more importantly, by its acute pro-inflammatory properties,¹⁴⁻¹⁶ as illustrated by its use in a preclinical model of paw edema in rats. In contrast, other carrageenan types exhibit a more moderate profile. Moreover, the anticoagulant properties of λ-CAR may limit its applications due to the potential risk of internal bleeding, particularly in oncological contexts where patients may have a fragile vascular system,¹⁷ unless anticoagulant or hemocompatibility properties are specifically desired.¹⁸ In view of these aspects, the use of carrageenans—and more specifically of the λ-type—in the field of nanomedicine is still in its early stages. Moreover, most preliminary studies have also preferred κ-CAR and ι-CAR types due to their convenient gelling properties, making them more suitable than λ-CAR for building hydrogels or self-assembled polymeric NP. Overall, the applications tend to focus more on leveraging the rheological properties of these polysaccharides rather than their bioactivities of interest. In fact, only a few articles explicitly mention λ-CAR as a framework for biomedical NP, and these initial attempts do not mention in vivo experiments, as they encountered the same limitations of innocuity as described above.¹⁹⁻²³ Additionally, many marine polysaccharides such as Fucoidan are increasingly being studied as macromolecule-based chelates for radioisotopes or lanthanides, unlike λ-CAR which remains largely unaddressed in this domain.^24,25

Controlled depolymerization of λ-CAR into oligosaccharides offers a promising solution to these issues^26,27 and a way to unlock their tremendous potential as biofunctional coatings for NP. Indeed, these low-molecular-weight derivatives (typically ranging from 0.5 to 10 kDa) demonstrate significantly better physicochemical features for systemic administration than their polysaccharides counterparts, including enhanced solubility, bioavailability and reduced viscosity. Furthermore, these transformations can enhance therapeutic efficiency, discarding undesirable biological properties of polysaccharides while retaining the desired bioactivity of interest. In this regard, several promising oligosaccharides derived from λ-CAR (λ-COS) that display better druggability have already been proposed for various pathological contexts, such as oncology or infectiology.^7,28 Despite this research, surprisingly, only a few studies have reported the utilization of oligosaccharides from carrageenans as ingredients in NP formulations, regardless of their type (κ, ι, λ).²⁹⁻³³ This is especially true for λ-COS, for which, to the best of our knowledge, none have yet been described in the literature for applications in nanomedicine.

Consequently, the present study aimed to determine whether a biofunctional λ-COS framework can be applied to build novel NP-based structures and impart colloidal stability or chelation, a suitable balance of biological properties, and beneficial pharmacokinetic properties. To achieve this, λ-COS-based NP (λ-COS NP) were prepared with extremely small encapsulated Mn-doped iron oxide (ESIONP) cores, favored for their versatile properties as contrast agents in magnetic resonance imaging (MRI) to facilitate prospective biodistribution studies.³⁴⁻³⁶ Next, a thorough physicochemical characterization of this new probe was performed, revealing a second biofunctionality of the λ-COS framework, which behaves as a chelate of Mn²⁺ ions that can partially uncouple in the plasma. Importantly, the absence of common adverse bioactivities attributable to native λ-CAR was confirmed in the case of λ-COS NP. Therefore, we aimed to evaluate the input of this new sugar scaffold for NP in vivo applications by conducting MRI-based studies in mice at low and high fields.

2. Experimental Section

2.1. Synthesis of λ-COS and Basic Physicochemical Characterisations

A 200 mL solution of native λ-CAR (5 mg/mL) in Milli-Q water was prepared, purged with argon flow, and left to solubilize overnight. Next, 4.5 mL of H₂O₂ (30%, Merck, NJ, USA) was added, and the reactor was sealed. The mixture was then placed in a heating incubator at 60 °C and stirred at 200 rpm. The depolymerization was monitored by collecting aliquots every hour to estimate the molecular weights by size-exclusion high-performance liquid chromatography until it reached an M_n of 3.8 kDa (M_w was estimated at 7.2 kDa, yielding a dispersity of the polymer of 1.9) (corresponding to approximately 30 h of reaction).

The molecular weights of the final λ-COS were estimated by size-exclusion high-performance liquid chromatography using an Agilent 1260 Infinity II system (Santa Clara, CA, USA) composed of two analytical columns (TSK-GEL G4000PW and TSK-GEL G3000PWXL; Tosoh, Japan) coupled with a Refractive Index Detector. The samples were eluted at 25 °C using a solution of NaNO₃ (0.1 M) at a flow rate of 0.8 mL·min^–1. Molecular weights calibration was performed using pullulan standards (from 10 to 806 kDa, Polymer Standards Service GmbH, Germany) for molecular weights higher than 10 kDa, and heparin standards (from 1200 to 5200 Da, Iduron, UK) for molecular weights below 10 kDa. The number-average molecular weight (M_n), weight-average molecular weight (M_w), dispersity (I), and degree of polymerization (DP) were calculated using a previously published procedure³⁷ and the following equations

1

2

3

4

where N_i is the number of moles of polymer species, M_i is the molecular weight of the polymeric species, and M₀ is the molecular weight of the monomeric unit estimated at 563.2 Da (C₁₁H₁₅O₂₀S₃).

The degree of sulfate was determined by Azure A colorimetric assay, as previously described.²⁶ Briefly, in a 96-well plate, 20 μL of sample with concentration ranging from 0 to 0.03 mg mL^–1 was incubated at 37 °C for 15 min with 200 μL of (7-aminophenothiazin-3-ylidene)-dimethylazanium chloride solution, named Azure A (Merck, NJ, USA) at 0.01 mg·mL^–1. Absorbance (BMG Labtech FLUOstar Omega, Germany) was read at 630 nm, and the degree of sulfation was determined using a calibration curve built from dextran sulfate, with a known sulfur content of 18.1% (w/w) (Merck, NJ, USA). The degree of sulfate of λ-COS was 22% (±3%) w/w sulfate group.

2.2. Synthesis of λ-COS NP

First, λ-COS (40 mg) was solubilized in deionized water (1 mL). Then, 30 μmol of MnCl₂·4H₂O (Sigma-Aldrich >99%) was dissolved in 1 mL of 0.05 M HCl and 37 μmol of FeCl₃·6H₂O (Sigma-Aldrich >99%) in 0.25 mL of deionized H₂O. In a 10 mL microwave tube, the λ-COS solution was added, followed by the Mn and Fe solutions. Next, 0.25 mL of hydrazine monohydrate (64–65%, Merck, NJ, USA) was carefully added before rapidly sealing the tube and placing it in the microwave unit (Discovery 2.0, CEM, NC, USA). The mixture was heated at 70 °C under magnetic stirring for 3.45 min. The reaction mixture was purified by gel filtration through a PD10 desalting column (GE Healthcare, IL, USA) using Milli-Q H₂O as the eluent.

2.3. Physico-Chemical Characterizations of the λ-COS NP

2.3.1. Dynamic Light Scattering Measurements

The hydrodynamic diameter and ζ potential were measured in low-volume polystyrene cuvettes and folded capillary cells, respectively, using a Ζetasizer Ultra (Malvern Instruments, UK). Measurements were performed in triplicate with 10 s of equilibration at 25 or 37 °C. The stability of the NP over time was also assessed in different media, including saline (NaCl 0.9%), Phosphate Buffer Saline (PBS, pH = 7.3, containing NaCl, KCl, Na₂HPO₄ H₂O, KH₂PO₄) and Dulbecco’s modified medium (DMEM; Gibco, Fisher Scientific, NH, USA) cell culture medium was also assessed.

2.3.2. Inductively Coupled-Mass Spectrometry (ICP–MS)

Ten μL of the sample was diluted in 1 mL of 2% HNO₃ Optima solution and left to react overnight. The elements were dosed using an iCAP-Q (Thermo Fischer).

2.3.3. Thermogravimetric Analysis

The samples (λ-COS and λ-COS NP) were frozen overnight at −80 °C and subsequently freeze-dried at room-temperature under vacuum with COSMOS20K-Cryotec (France). The resulting powder (5–20 mg) was then subjected to thermogravimetric analysis, heating from 20 to 1000 °C at a rate of 10 °C·min^–1, under argon flow of 20 mL·min^–1, on a TGA/DSC3+ instrument (Mettler Toledo, OH, USA).

2.3.4. Azure A Dosage

The coating was quantified using the Azure A dosage described in Section 2.1, with λ-COS as the sulfur standard.

2.3.5. Transmission Electron Microscopy (TEM) and Dispersive X-ray Spectroscopy (EDX)

TEM images were obtained using a 2100F TEM (JEOL, Japan). The gun was a Schottky emitter with a high voltage of 200 kV. The grid was observed with a Gatan Rio16IS Camera in TEM mode. STEM Imaging was conducted using a JEOL ARM200 Cold FEG spherical aberration-corrected TEM equipped with a cold field emission gun operated at 200 kV. EDS spectrum and mapping were performed using an EDX SDD CENTURIO-X instrument (JEOL). All STEM images were obtained using the BF and HAADF STEM detectors. A drop of a diluted NP solution was drop-cast onto a copper grid covered with an amorphous carbon film. The grid was degassed overnight under a secondary vacuum.

2.3.6. Magnetic Relaxometry

The relaxation times T₁ and T₂ were measured (in triplicates) with mq60 minispec (Bruker Biospin GmbH, 1.5T, Bruker, MA, USA) by testing 300 μL of several concentrations ranging from 2 to 20 mM (Fe + Mn) at 37 °C, with Milli-Q H₂O as the solvent. Standard T₂ Carr–Purcell Meiboom Gill and T₁ inversion—recovery spin–echo sequences were used to perform the measurements. The longitudinal (r₁) and transverse (r₂) relaxation properties were determined by plotting the 1/T₁ and 1/T₂ values in s^–1 against the (Fe + Mn) concentration in mM, with Milli-Q H₂O as a blank. The relaxation values r₁ and r₂ in s^–1·mM^–1 were derived from a linear fit of the resulting curve corresponding to the slope.

2.3.7. Raman and Infrared Spectroscopy

Raman spectra were acquired on freeze-dried samples (same protocol as described in Section 2.3.3) using a Jobin Yvon High-Resolution Raman spectrometer (LabRAM HR-Evo, LabRAM, UK) equipped with a microscope (Olympus BX 41) and a Peltier-based cooled charge-coupled device detector (CCD). A laser diode was used as the excitation source at a wavelength of 532 nm. Spectra were recorded using the acquisition LabSpec software at RT, with a resolution of 0.4 cm^–1. Infrared spectra were acquired using a UATR Two FT-IR spectrometer, LiTaO3, with an MIR detector (PerkinElmer, MA, USA), from 4000 to 450 cm^–1 with 64 scans, a nominal resolution of 2 cm^–1, and a scan speed of 0.2 cm·s^–1. Curves were processed using LabSpec and baseline was corrected.

2.3.8. X-ray Photoelectron Spectroscopy (XPS)

XPS experiments were performed using a Versaprobe III Physical Electronics (ULVAC, MA, USA) spectrometer with a monochromatic X-ray source (Al Kα line of 1487 eV) calibrated using the 3d_5/2 line of Ag (368.3 eV). The samples were mounted on nonconductive tape. Z-alignment was performed to determine the optimal sample height before each measurement. Argon Ion Gun and E-neutralizer were used for charge neutralization. Elemental quantification was performed on a survey scan with a step energy of 0.5 eV and pass energy of 224 eV. High-resolution regions were acquired with a step energy of 0.05 eV, pass energy of 55 eV and time per step of 50 ms. Data were analyzed using CasaXPS software (2.3.25 PR 1.0). The spectra were calibrated according to the sp³ C–C peak, and the BE was fixed at 284.8 eV.

2.4. Bioactivities Evaluations and Cell Culture Experiments

2.4.1. Cell Culture Conditions

Cell lines were purchased from ATCC (American Type Culture Collection, France), and human embryonic kidney cells (HEK293) and murine macrophages (RAW264.7) were chosen as models for healthy cells and immune cells, respectively. Cells were cultured in DMEM supplemented with 10% decomplemented FBS and 1% penicillin/streptomycin (both from Gibco, Thermo Fisher Scientific, MA, USA). All cell lines were maintained at 37 °C in a humidified atmosphere containing 5% CO_2, and were mycoplasma-free (MycoAlert Mycoplasma Detection Kit).

2.4.2. Evaluation of Inflammatory Activity

In a 24-well plate, 1.8 × 10⁵ cells·well^–1 were treated with different samples (λ-CAR, λ-COS, or λ-COS NP) at a concentration of λ-CAR or λ-COS (assessed by Azure A dosage for the λ-COS NP) fixed at 100 μg·mL^–1 for 24 h of incubation. IL-6 and TNF-α levels in the cell supernatants were quantified using ELISA Development Kits (PeproTech), according to the manufacturer’s instructions.

2.4.3. Anticoagulant Xa/IIa Inhibitory Activity

The inhibition of Xa and IIa have been assessed by a colorimetric assay using antithrombin (ATIII), Thrombin (IIa), Factor Xa and the corresponding chromogenic substrates Xa/IIa (STACHROM AT III and STACHROM HEPARIN kits from Stago, France). In brief, 25 μL of the sample (λ-CAR, λ-COS, or λ-COS NP, performed in triplicate) or Milli-Q H₂O (as blank) were mixed in a 96-well plate with 25 μL of ATIII (0.625 μg·μL^–1 in deionized water) and incubated at 37 °C for 2 min. Then, 25 μL of Factor Xa or Thrombin IIa (11.25 nkat·mL^–1 in deionized water) was added and incubated for 2 min at 37 °C. Finally, 25 μL of the corresponding chromogenic substrate (Xa or IIa) was introduced, and the absorbance was read at 405 nm in kinetic mode for 4 min (BMG Labtech FLUOstar Omega). The initial velocity, corresponding to the slope of the absorbance with respect to time in h^–1, was calculated. The Inhibition percentages were calculated using the following equation

5

where Vi(blank) and Vi(sample) are the initial velocities of the blank and sample, respectively. Data were processed with Graph-Pad Prism 8 and IC₅₀ were calculated by first normalizing the values and fitting to the model “absolute IC₅₀,” except for IIa inhibition of the λ-COS NP because the values did not follow an IC₅₀ inhibition curve.

2.4.4. Metabolic Activity Assay

In a 96-well plate, cells were seeded at 10⁴ cells·well^–1 in 200 μL of adequate cell culture medium and incubated for 24 h. The medium was then replaced with 200 μL of fresh medium containing native carrageenan, λ-COS, or λ-COS NP at 0, 10, 50, 100, or 200 μg·mL^–1. After 24 h of exposure, 25 μL of 3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide (MTT) solution was added to each well, followed by 3 h of incubation. The supernatant was then removed and replaced with 100 μL of DMSO to dissolve the formed formazan crystals, and the absorbance was read at 540 nm. Viability was calculated using the following equation

6

2.4.5. Cellular Uptake of Nanoparticles

Cells were seeded in 24-well plates (10⁵ cells·well^–1). After 24 h, cells were treated with 100 μg·mL^–1 native carrageenan, λ-COS, or λ-COS NP λ-COS NP. After 24 h of incubation, λ-COS NP internalization was qualitatively assessed by microscopic observation (ZEISS microscope, Axio Observer Z9, magnification: ×10, ZEISS, Germany) using a Prussian blue nuclear fast red staining kit (Abcam). Images were constructed using Fiji software.

2.4.6. Statistical Analysis

One-way or Two-way analysis of variance (ANOVA) and t tests were performed to identify significant differences between the control and experimental groups. All experimental data were obtained from at least three independent experiments. All statistical analyses were performed using GraphPad Prism version 8 for Windows. A probability (p) value < 0.05 was considered statistically significant.

2.5. In Vivo Experiments

2.5.1. Animal Ethics

All animal experiments were performed in accordance with the Spanish policy for animal protection (RD53/2013), which meets the requirements of Directive 2010/63/UE. The 7T procedures were approved by the Ethical Committee of the CIC biomaGUNE and authorized by the local authorities (PRO-AES-SS-225).

2.5.2. MRI 7-T and 1-T In Vivo Imaging

The λ-COS NP solution was concentrated with a 30 kDa filtration unit (Amicon, Merck) for 10 min in a centrifuge (Minispin, Eppendorf, Germany) at 7000 rpm, resuspended in saline (NaCl 0.9%), placed in an ultrasound bath for 10 min, and filtered through a 0.22 μm filter (Merck Millipore). Fifty to one hundred microlitres of this solution were intravenously injected into 7 week-old BALB/cJRj mice (Janvier Lab, France). Biodistribution studies were assessed under two distinct magnetic field conditions.

MRI 1 T was performed at the ICTS-BioImaC of the Universidad Complutense, node of the ICTS ReDIB (https://www.redib.net/), using a 1 T benchtop MRI scanner (ICON 1 T-MRI; Bruker BioSpin GmbH, Ettlingen, Germany). The system consisted of a 1 T permanent magnet with a gradient system capable of supplying a gradient strength of 450 mT·m^–1. A solenoid mouse-body RF coil was used. Animals (6 week-old female Balb/c mice) were anaesthetised using 2% isoflurane (IsoFlo, Zoetis, NJ, USA). The main MRI experiment consisted of three-dimensional T₁ weighted images used to monitor the evolution of contrast enhancements, before and after λ-COS NP intravenous injection. The isotropic T₁ weighted experiments (0.5 × 0.5 × 0.5 mm) were acquired using a gradient echo sequence with a repetition time of 20.3 ms, an echo time of 2.0 ms and flip angle of 20°. The total acquisition time was 6 min.

Other parts of the experiments were performed at the ICTS-CIC BiomaGUNE on a 7 T Bruker Biospec 70/30 USR MRI system (Bruker Biospin GmbH, Ettlingen, Germany), interfaced with an AVANCE III console. A BGA12 imaging gradient (maximum gradient strength 400 MT·m^–1) system, with a 40 mm diameter quadrature volume resonator, was used for MRI data acquisition. Anaesthesia was induced using 4–5% isoflurane in 30% oxygen and maintained at 1–2% isoflurane throughout the experiment. The animals were positioned in a customized 3D-printed bed equipped with a head holder and maintained at 37 °C using heated air pumped through an MRI-compatible system, interfaced with Monitoring and Gating model 1030 (SA instruments, NC, USA). Anal. temperature control and respiration (monitored through a respiratory pad) were recorded throughout the experiment. Anatomical images of the abdomen were acquired with a Bruker gradient echo FLASH sequence using the following parameters: TE 4 ms, respiration synchronized (TR 600 ms), flip angle 30°, 2 averages, 256 × 256 points, a field of view of 4 cm × 4 cm and 12 noncontiguous slices with a slice thickness of 1.0 mm. Quantification was performed using T₂ maps to facilitate comparison on different days of the experiment. T₂ maps were obtained using a multislice spin–echo (MSME) sequence. The TE values varied in 20 steps, ranging from 8 to 160 ms, and were synchronized with respiration (TR 4000 ms). T₂ maps were created with a customized python-based script to adjust and calculate the maps and were analyzed using the ITK-SNAP software. Images were acquired before injection and between 30 min, 2 and 24 h after injection of the NP solution.

2.5.3. Ex Vivo Analysis

Histopathological analyses were performed by fixing the tissues in Formalin 4%. This was followed by a series of washes with ethanol (50%, 75%, and 100% concentrations) and xylene and then embedded in paraffin. Subsequently, the tissues were sectioned using a microtome. Two different staining methods were applied: (i) hematoxylin and eosin staining to observe the integrity of the tissues, and (ii) Prussian blue staining to highlight the presence of iron in the cells. Final observations were performed using a confocal microscope (Zeiss LSM 880 Airyscan). Organs were freeze-dried according the procedure described in Section 2.3.3. Using Christ, Alpha 2-4 LSCplus (Germany). The Mn and Fe contents in the freeze-dried organs were quantified after digestion in a 2% HNO₃ Optima Microwave Digestion System Speedwave XPERT (Berghof SpeedWave Xpert, Germany) using an established protocol for meat digestion. The elements were dosed as described in the ICP–MS section.

2.6. Data Curation

All data and graphs were processed using Excel (Microsoft, USA) or GraphPad Prism (release 8 for Windows), and the schemes were implemented using Biorender (Canada) and PowerPoint (Microsoft, USA).

3. Results and Discussion

3.1. Synthesis of λ-COS NP

λ-COS was obtained from native λ-CAR polysaccharides using a previously published, H₂O₂-based radical depolymerization method.³⁸ The basic physicochemical parameters of λ-COS include an M_n of 3.8 kDa with a dispersity of the polymer of 1.9 (measured by size-exclusion high-performance liquid chromatography with spectra displayed in Figure S1), and a degree of sulfate of 22 (±3) % (measured by a colorimetric quantification assay). λ-COS-based hybrid NP embedding free Mn²⁺ ions and extremely small iron oxide (ESIONP) cores, referred to here as λ-COS NP, were prepared by a one-pot synthesis divided into two steps (Figure 1).³⁹ First, manganese dichloride tetrahydrate (MnCl₂·4H₂O) and iron trichloride hexahydrate (FeCl₃·6H₂O) precursors (molar ratio of 1:1.2, Mn) were added together to a λ-COS solution (40 mg·mL^–1), forming a water-soluble metals- λ-COS coordination complex. The ability of Mn²⁺ and Fe³⁺ to bind with carrageenans polysaccharides through interactions between the metal ions and the negatively charged sulfate (−SO₃^–) groups,⁴⁰⁻⁴² and to a lesser extent hydroxyl (−OH) groups,^19,40,43 has already been documented in several articles. However, this is the first time such effects have been observed for OS. Dynamic light scattering measurements showed no aggregation or self-assembled NP at this stage, most likely due to the low-molecular-weight of oligosaccharides (limiting the number of available binding sites for cross-linking compared to native PS). And yet, effectiveness of the water-soluble metals- λ-COS coordination complex was corroborated by other experiments (Figure S2). First, an FTIR analysis of the free λ-COS and the metals- λ-COS coordination complex was carried out. Participation of hydroxyl (−OH) groups in the coordination of metals was revealed by a broadening of the band in the complex’s spectra, compared to the one obtained from the free λ-COS. A clear down-shift of the asymmetric S = 0 stretch from ∼1220 to ∼1209 cm^–1 was detected in the complex’s spectra, confirming the participation of sulfate groups in the stabilization of the metal ions. Moreover, ζ-potential measurements of the λ-COS solution showed a slight shift from −13.9 to −11.3 mV upon addition of the metal salts. Finally, the peak at 573 cm^–1, characteristic of λ-COS skeleton’s additional features, displayed a higher intensity after complexation. This could be compatible with the presence of an additional band corresponding to the Metal–oxygen (M–O) bond vibrations, which—shifted from 533 to the 573 cm^–1 peak, when metal salts were alone. Next, UV–visible analyses were performed to complete the picture and higher absorbances were observed in the UV region (200–300 nm^–1) of the metals- λ-COS coordination complex’s spectra, compared to those of free metals and λ-COS taken alone, indicating a ligand-to-metal charge transfer consistent with a complex formation (Figure S2).⁴⁴

Figure 1.

Synthesis of λ-COS NP. λ-COS was prepared by a radical-based depolymerization method and next λ-COS NP were prepared using a two-step method: 1) Mn + Fe λ-COS coordination complex and then in situ 2) crystal growth of ESIONP using reductive thermal treatment.

In the second step, using a previously developed method for preparing various natural PS- and OS-coated ESIONP,^45,46 hydrazine was added in situ, followed by a quick, mild thermal treatment (70 °C for 3 min 45 s). This process induced the rapid formation of intermediate iron hydroxides, as well as the partial reduction of Fe³⁺ into Fe²⁺, facilitating the nucleation and growth of crystalline ESIONP cores, which in turn drives a nanosized λ-COS matrix assembly as a stabilizer. The final hydrodynamic diameter (HD) of λ-COS NP, measured by dynamic light scattering, was 79 ± 2 nm. The size range of 50–150 nm is generally considered suitable for achieving prolonged circulation time (avoiding rapid clearance by the renal and mononuclear phagocyte systems), efficient cellular uptake, and passive targeting, particularly in antitumor applications.⁴⁷ The ζ-potential of λ-COS NP, measured at physiological pH, was highly negative at −21.0 ± 0.8 mV due to the presence of negatively charged sulfate groups in the λ-COS framework. This strong negative value is expected to provide effective electrostatic stabilization, which is crucial for preventing aggregation and ensuring colloidal stability.⁴⁸ A control reaction, consisting in the single addition of iron precursor without Mn²⁺, resulted in NP with a larger HD of 215 ± 3 nm and embedded low-quality amorphous ESIONP cores (data not shown), indicating the importance of Mn²⁺ addition in modulating the kinetics of intermediate iron hydroxide formation. By competing with the Fe binding sites, Mn²⁺ probably slowed the burst nucleation (i.e., iron hydroxides forming too rapidly which prevent the ordered crystal growth) and an iron hydroxide-induced higher aggregation of the λ-COS matrix.

3.2. Advanced Characterization of λ-COS NP

To further describe the final λ-COS NP, final amounts of Mn and Fe in the structure were quantified by ICP–MS (Figure S3). The final loading efficiencies of the metals in the NP were approximately 20% of the initial amounts introduced in both cases. Subsequently, the λ-COS NP were resuspended in an aqueous solution of Zn²⁺ at 2.5 mM, as Zn²⁺ can quantitatively substitute divalent cations that are complexed within organic structures.⁴⁹ After ultracentrifugation of the solution using a 3 kDa cutoff, Mn and Fe were quantified in the filtrate. Results showed no detectable Fe, indicating that all the iron complexed at the first step had reacted and was finally contained within the encapsulated ESIONP cores. On the other hand, 90% of the total Mn was released, which corresponds to the complexed Mn inside the λ-COS NP, while the remaining 10% were Mn that diffused in ESIONP cores during crystal growth in the second step of the reaction. Consequently, the ESIONP composition could be refined to the following stoichiometry: Mn_0·2Fe_2·8O₄ and will now be referred to as ferrite cores.

As shown in Figures 2A and S4, large field-of-view TEM images in bright-field and dark-field modes highlighted these inorganic cores of round-like shape arranged in bunches among the λ-COS NP organic matrix. Core diameters computed from these images exhibited a distribution of 2.6 ± 0.4 nm on average (Figure 2A). This, in relation to the HD of 79 ± 2 nm, suggests the presence of a cohesive λ-COS matrix encapsulating multiple cores, rather than a simple surface coating. EDS maps from the TEM image also indicated colocalization between the Fe atoms and the observed cores, while Mn atoms were distributed more largely due to the Mn population complexed in the λ-COS framework (Figures S5 and S6), in addition to the Mn doping found in the cores. Additional HRTEM microscopy analysis revealed their good crystallinity, as confirmed by the numerous lattice fringes that were detectable (Figure 2B). The measured d-lattice spacings from the SAED patterns were in good agreement with the ones found for spinel manganese ferrite, including 2.38, 2.14, 1.46, and 1.17 Å for the (220), (311), (400) and (331) planes, respectively (Figure S7).^50,51 Finally, Raman spectroscopy is a valuable tool for characterizing the crystallographic structure of these encapsulated ferrite cores.^52,53 In the spinel arrangement, the oxygen anions are packed in a face-centered cubic structure, forming a basic structure, and tetrahedral (Td) and octahedral (Oh) sites can be occupied by metallic ions to form normal, mixed, or inverse structures. Indeed, in normal spinel structures, divalent ions reside in the Td sites, while trivalent ions are in the Oh sites. In the inverse structure, however, such as in magnetite (Fe₃0₄), the trivalent ions are equally distributed between the Td and Oh sites, with divalent ions in the Oh sites. Since Mn²⁺ can occupy both Td and Oh sites, Mn doped-Fe₃O₄ typically build mixed structures with varying degrees of inversion, depending on the ratio of Mn-substitution, the type of synthesis, or the particle size.^52,54,55 The pattern obtained with λ = 532 nm excitation closely matched this scenario, displaying an intense peak at 614 cm^–1, in particular. This is attributed to the A_1g mode and corresponds to the Fe³⁺–O stretch vibration in the Td sites, with shifts due to the local distortion induced by the Mn-substitution (Figure 2C).⁵⁴ The broadening of this band toward higher wavenumbers specified the vibration modes of other cations in the Td sites, prone to be Fe²⁺ or Mn²⁺ ions and indicative of partial inversion in the spinel. The other peaks below 600 cm^–1 are associated with the vibration between the cations located at the Oh sites and oxygen groups. Accordingly, the bands observed at ∼320 and ∼450 cm^–1 can be considered as one of the E_g (symmetric bending) and T_2g (antisymmetric bending) modes of Fe³⁺, with shifts that are more consistent with the normal characteristics of the spinel.^52,54,55 However, the broadening of the T_2g bands can be attributed to the presence of other types of ions (Mn²⁺ or Fe²⁺) at the Oh sites.

Figure 2.

Characterization of λ-COS NP. (A) TEM images and size distribution, (B) HRTEM and d-spacing and (C) Raman spectra at 785 nm of the encapsulated ferrite cores. (D) Thermogravimetric analysis curves of the λ-COS and λ-COS NP.

Next, the proportions of the different elements constituting the λ-COS NP were investigated. Colorimetric quantification of the sugar-based coating,²⁶ when compared to ICP–MS analysis of the overall metal ions, yielded a final λ-COS to metal (Fe + Mn) w/w ratio of 11.1. A thermogravimetric analysis was also performed to characterize the balance between organic coating and ferrite cores amounts. As shown in Figure 2D, the degradation of λ-COS NP was divided into three stages: an initial loss of up to approximately 200 °C, corresponding to the evaporation of bound water; a second step from 200 to 600 °C, attributed to the degradation of the organic coating; and finally, a plateau representing the mass retention of the ash content and metal oxide formation that accounted for 31.1% of the weight. A comparison of this plateau with that obtained from the thermogravimetric analysis of the free λ-COS (24.2% remaining weight) indicates that the encapsulated ferrite cores of λ-COS NP accounted for approximately 7% of the weight of the sample. The weight ratio of λ-COS/(Fe + Mn) w/w (13.3) obtained in this experiment was consistent with the ratio estimated earlier (11.1). Next, multiangle dynamic light scattering measurements were used to estimate NP concentration in the solution. Based on previous data obtained and the M_w of λ-COS, several parameters were estimated, namely, the average number of oligosaccharide chains, ferrite cores and complexed Mn per total NP, to determine whether the orders of magnitude were consistent with the described system (Table S8). The calculations indicated that a single NP contained approximately 1.3 × 10⁶ λ-COS chains, 2.7 × 10⁴ ferrite cores, and 5.7 × 10⁶ Mn atoms, which aligns reasonably with earlier descriptions of the NP. More explicitly, these values correspond to roughly 4 Mn atoms per λ-COS chain and about 50 λ-COS chains per ferrite core (equating to a density of 2.34 chains/nm² of ferrite cores’ surface), in good agreement with previous data.^40,45

3.3. XPS and FTIR Characterization of λ-COS NP

To complete the chemical characterization of λ-COS NP, XPS analysis was performed on both λ-COS NP and λ-COS (Figures 3A–E and S9). The organic elements and chemical functional groups expected in λ-COS/CAR were retrieved from the NP spectra with C, O, S, and Na as the counterions (Figure 3A). A minor proportion of nitrogen (3.7%) was detected in the free λ-COS and in the NP samples, which was unexpected considering the theoretical chemical structure of the sugar. The C/S ratio in the λ-COS NP sample (3.8) was significantly lower than that in the free λ-COS coating (11.6), indicating a partial desulphation during the NP synthesis. Indeed, in the C 1s spectra, deconvolution attributed to C–O also decreased (Figure S9) and that might correspond to this desulphation.⁵⁶ The asymmetric O 1s peak in the λ-COS NP XPS spectra can be deconvoluted into various contributions (Figure 3B). The lowest binding energy (530.5 eV) is commonly associated with metal–oxygen bonds, while the higher contributions (up to 531 eV) correspond to the oxygen bonds present in the saccharidic coatings (C–O –SO^3–, C=O). It is interesting to note a shift in the O_ads binding energy toward higher values, compared to the spectra of the free λ-COS, indicative of the change in their neighboring environment resulting from the participation of the hydroxyl and sulfate groups in metal stabilization.^56,57 In addition, adsorbed water in the free λ-COS sample, detected through the signal at 536–537 eV, was no longer present in the NP sample. Regarding the metallic ions, two peaks were observed at 642 and 654 eV for Mn, corresponding to Mn 2p_3/2 and Mn 2p_1/2, respectively (Figure 3C). The presence of the satellite peak at 645 eV (shakeup) confirmed the presence of paramagnetic Mn²⁺ complexed in the nanoarchitecture.⁵⁸ Deconvolution of the Mn 2p_3/2 could indicate the minor presence of other higher oxidation forms of Mn (+ III or + IV).,⁵⁹ and yet signals from the Mn 3s were split into two peaks at 83.76 and 89.85 eV, giving a ΔE of 6.1 eV (Figure 3D). This ΔE value indicated a prevalent Mn²⁺ oxidation state, as reported in the MnO XPS spectra.⁵⁹ On the other hand, analysis of Fe signals is entirely related to the ferrite cores. This can be seen from Figure 3E: the two main asymmetric peaks, Fe 2p_1/2 and Fe 2p_3/2, with binding energies at 710.5 and 723.8 eV, giving a doublet spacing of 13.3 eV. The presence of their respective satellite peaks at approximately 715 and 728 eV reveals that Fe³⁺ is the main oxidation state present in the core, even if subtle shoulder can indicate the minor presence of Fe²⁺. Several contributions to Fe 2p_3/2 might be detected, potentially reflecting the three different environments of the Fe³⁺ ions. The apparent shift toward the higher binding energy at 711 eV suggests the predominance of Oh site occupation and hydroxyl group binding, which is consistent with the Raman results (Figure 2C).

Figure 3.

Coating characterisations; (A) table of atomic quantification, XPS spectra of (B) O 1s, (C) Mn 2p, (D) Mn 3s, (E) Fe 2p and (F) FT-IR spectra of λ-COS NP.

Next, FTIR analyses were performed in an attempt to confirm and decipher the chemical groups of λ-COS that interact with the different elements (Mn²⁺, ferrite cores) in the final NP (Figure 3F). The main peaks corresponding to the λ-COS’s functional groups were successfully retrieved in the two spectra. The band corresponding to the hydroxyl groups (3600–3300 cm^–1) was slightly broader in the λ-COS NP spectra, confirming their participation in the structure stabilization. The S=O asymmetric stretch shifted from 1220 cm^–1 (free λ-COS spectra) to 1209 cm^–1 in the final NP, together with a change in relative intensities of the S=O symmetric stretch (1010–1035 cm^–1). This is indicative, once again, of the importance of sulfate groups in the coordination of the Mn²⁺ and ferrite cores. Glycosidic linkage (1070 cm^–1), anhydro-galactose ring (930 cm^–1) and C–O–S peaks (845 cm^–1) remained unchanged. To be noted, the signal at 1737 cm^–1 in the λ-COS spectra, which can be attributed to the open reducing ends exposing an aldehyde group, were not retrieved in the one of λ-COS NP. As this change was not detected in the metals–λ-COS coordination complex’s spectra displayed previously, it could indicate an additional contribution of the reducing ends toward the ferrite core’s stabilization.

3.4. Magnetic Characterization of λ-COS NP

All these structural features strongly influence the magnetic properties of the λ-COS NP. To study them, magnetometric measurements were performed. First, the magnetization, according to magnetic fields (M(H)) recorded at 10 K of the lyophilized samples, revealed that the sample was not completely saturated in the field range used (4 MA·m^–1), thus confirming the presence of complexed paramagnetic ions and/or very fine NP (Figure 4A).⁶⁰ Previous literature for extremely small ferrites often reports the presence of a magnetic dead layer in which spin canting and surface magnetic disorders profoundly impact saturation magnetization by reducing the net magnetic alignment.⁶¹ It was observed that, up to 150 K, the M(H) curves displayed full paramagnetic behavior (Figure S10). Further to this, the temperature dependence of the magnetization was measured with the ZFC-FC procedure from 2 to 270 K under a magnetic field of 8 kA·m^–1 (Figure 4B). The derivative of the temperature difference between the FC and ZFC branches provided a very small experimental blocking temperature (T_b) of 3 K (graph not shown). Next, all ZFC-FC and M(H) curves were fitted to a model of noninteracting particles (labeled “Total Fit” in Figure 4A,B). Both, ZFC-FC and M(H) at 10 K curves fit quite well to this model by assuming a superposition of two distributions of populations (identified as dist1 and dist2, respectively), given by two different supermoments of approximately 150 Bohr magnetons (μ_B) (1.3%) and 4 μ_B (98.7%). In this case, the small bump, or relative maximum of the ZFC-FC curve, located at 5 K is due to the small fraction of “dist 1” 150 μ_B NP. From the M(H) fit in Figure 4B, the saturation magnetization (Ms) of this NP superparamagnetic population can be estimated, which is approximately 26 A·m²·kg^–1 (without correction of the organic content). Also, whatever the value for the spontaneous magnetization of the particles used, it was possible to estimate the magnetic size of the different objects in the two populations. Taking M = 400 kA·m^–1, the first population magnetic size was estimated to be approximately 2.4 nm and the second was approximately 0.75 nm. For the first, this result matches well with the magnetic core diameter calculated for the ferrite cores population from TEM images (2.6 nm), and for the second, it reinforces the previous indications about the presence of complexed Mn in the λ-COS matrix.

Figure 4.

Magnetic characterization of λ-COS NP; (A) magnetization curve M(H) at 10 K (experimental moment) and (B) ZFC-FC curves (experimental moment) with theoretical fitting (total FIT) made from two distributions of populations (dist1 and dist2).

To assess the potential of the λ-COS NP as MRI CA, its relaxometric properties were measured at 37 °C using a 1.5 T relaxometer. Additionally, serial dilutions of phantoms in 1 and 7 T MRI scanners were also performed to study their response in low-middle and high magnetic fields, and both were used for upcoming in vivo studies (Figure S11). At 1 T, the relaxivity values r₁ and r₂ measured from the phantoms were 2 and 14.6 mM^–1·s^–1, respectively. These results were comparable to those obtained using the relaxometer with a close magnetic field of 1.5 T: r₁ of 2.47 ± 0.01 mM^–1·s^–1 and r₂ of 18.2 ± 0.2 mM^–1·s^–1. At 7 T, the r₁ value computed from the signals remained roughly similar (2.3 mM^–1·s^–1), while r₂ value was much higher 52.7 mM^–1·s^–1, indicating that λ-COS NP are more appropriate for T₂-weighted imaging at this high field. Explanations could lie in the fact that, at high fields, the T₁ reduction induced by complexed Mn ions in the NP frameworks is not strongly enhanced compared to low fields, while the T₂ contrast arising from the ferrite cores, also present in the NP, is highly favored.

3.5. Stability of λ-COS NP and Controlled Release of Mn in Serum

The high mass ratio of the coating to the nanoparticle core, combined with the highly negative ζ potential, confers high colloidal stability to the NP at physiological pH. To go further, the stability of λ-COS NP across a pH range of 3.5–9.5 was also studied by measuring the HD and ζ potentials (Figure 5A,B). The HD remained almost constant, ranging between 71 and 80 nm (as measured by recording the signal intensity which is the most sensitive criterion used to detect the eventual presence of aggregates), with the minimum observed at a pH of 7–7.3, corresponding to the typical physiological pH. The ζ potential remained stable at approximately—–30 mV until pH 8.5, beyond which it decreased to −41 mV under basic conditions. A similar curve pattern was observed for the λ-COS alone, with a drop-off at a pH of 8.5 which is coherent with the previous literature.⁶² The stability of the NP over time in various media, relevant to future experiments, was also assessed by HD monitoring (Figure 5C). In DMEM, a small increase in HD was observed, likely due to the presence of proteins that cross-link or adsorb onto the NP surface. However, the NP remained stable over 4 days, indicating the suitability of this medium for studying the effects of NP in cell-based studies. The results also showed that the HD was similar in saline (NaCl 0.9%) and Milli-Q H₂O, and it remained generally stable over 4 days. In contrast, aggregation and destabilization of the NP occurred in PBS, likely due to the high salt content. Therefore, saline was identified as an appropriate isosmotic buffer for further intravenous administration in in vivo mouse studies.^39,63

Figure 5.

Stability of the λ-COS NP in solution; (A) HD measured by dynamic light scattering (Z-average values) values according to the pH (between 3.5 and 9.5), (B) ζ potential according to the pH (between 3.5 and 9.5), (C) HD measured by dynamic light scattering (Z-Average values) of the λ-COS NP in different media (mQ H₂O, saline, PBS and DMEM) through time (between 0 and 4 days), (D) Mn release of the λ-COS NP in % in different media, (E) T₁ and T₂ measurements of the λ-COS NP in different media. Values were calculated as the mean of three measurements and SD.

Next, experiments were performed to assess Mn release in different media and assess its stability within the NP matrix (Figure 5D). For this, the λ-COS NP were ultracentrifuged using a cutoff of 3 kDa and Mn was dosed in the filtrate by ICP–MS. In mQ H₂O, Mn release did not exceed 6% even at 3 h, confirming they remain complexed in this medium. The experiment was also conducted in mouse plasma and, interestingly, a partial release of approximately 20% was observed that remained stable over time. This metal release from the NP in serum should modify its relaxometric characteristics and globally affect the r₁ value of the solution.^64,65 Indeed, it has been shown that free Mn²⁺, and its binding to protein carriers in the serum, greatly reduce the tumbling rates, leading to an increase of r₁. Relaxometric values were measured over time in mouse serum and compared to those obtained in mQ H₂O at the same concentration of 0.1 mg·mL^–1 in Fe + Mn (Figure 5E). As expected, due to the initial release of Mn²⁺, T₁ was immediately observed to be clearly lower in the mouse serum compared to mQ H₂O (34 ms against 160 ms, respectively). It then increased slightly until reaching a plateau of 58 ms at 1 h that is consistent with a re-equilibrium of the system. On the other hand, Mn²⁺ uncoupling from the NP and binding to protein carriers is not supposed to strongly affect T₂ shortening.⁶⁵ Indeed, despite T₂ being initially lower than in mQ water, probably owing to protein corona effects and stabilization of the suspension, it rapidly restored to a similar value of around 40 ms.

3.6. Innocuous Profile of λ-COS NP

Injection of λ-CAR is known to induce acute inflammation in vivo, promoting the recruitment of immune cells and expression of pro-inflammatory cytokines, which significantly limits its biomedical applications.^66,67 Multiple studies have demonstrated that λ-CAR activates macrophage responses via pattern recognition receptors, resulting in elevated levels of several cytokines, including tumor necrosis factor alpha (TNF-α), interleukin (IL) 6, IL-8, or IL-1β.^66,68,69 Depolymerization of λ-CAR to a very low molecular weight (<10 kDa), however, can reduce or suppress this pro-inflammatory activity.²⁷ As a pilot experiment to determine whether λ-COS NP display pro-inflammatory behavior, their effect on the secretion of TNF-α and IL-6—two key inflammatory markers-was tested in murine RAW 264.7 macrophages and compared to that of λ-CAR and free λ-COS. Before measuring cytokine levels, the impact of each candidate on RAW 264.7 cell viability was assessed over a concentration range of 0–200 μg·mL^–1 using an MTT assay to rule out cytotoxicity and ensure that any changes in cytokine expression were not due to reduced cell populations. The results confirmed that all tested compounds show generally good cell viability across different concentrations (Figure 6A), except for λ-COS and λ-COS NP at the highest dose tested (200 μg·mL^–1), where a slight, none-statistically significant effect were, observed.

Figure 6.

(A) Viability of RAW 264.7 macrophages measured by MTT after 24 h incubation with drugs, (B) TNF-α concentration and (C) IL-6 concentration in RAW 264.7 cell media 24 h after incubation with drugs (concentration of 100 μg·mL^–1 in sugar). Values represent mean ± SEM (n = 3), and p-values were calculated using one-way ANOVA with Tukey’s multiple comparison test. Anticoagulant activity: (C) IC₅₀ inhibition of Xa factor and (D) IIa factor (AT III = 0.625 μg·μL^–1 and factor Xa or IIa = 11.25 nKat.·mL^–1), (E) Sum-up of the different IC₅₀ activities assessed per drug. Values represent the mean ± SD of triplicate and were fitted through the “absolute IC₅₀” model after normalization to obtain the IC₅₀.

As expected, native λ-CAR treatment at 100 μg·mL^–1 triggered an inflammatory response, evidenced by a marked increase in TNF-α and especially IL-6 production (Figure 6B,C). In contrast, cells exposed to λ-COS and λ-COS NP maintained TNF-α and IL-6 levels close to untreated control cells. Overall, these findings provide initial evidence that the depolymerised derivative λ-COS, whether in free solution or as NP coating, does not elicit acute pro-inflammatory effects.

Another bioactivity of λ-CAR that can limit its applications is its moderate anticoagulant effect.²⁶ These effects have mainly been ascribed to either AT-III mediated inhibition of factors Xa and thrombin IIa, or to interaction with the heparin cofactor II involved in the coagulation cascade.⁷⁰ It has been proposed that λ-CAR follows a mechanism similar to heparin for the inhibition of Xa and IIa, which is effective through the activation of the endogenous modulator antithrombin III (AT-III).⁷¹ This mechanism involves the binding of a specific sequence among the polysaccharide to AT-III, leading to a conformational change-based activation that enhances its affinity for the ligand Xa and IIa. For factor IIa inhibition, a minimum chain length is required to directly provide an additional steric hindrance effect on this factor. Inhibition of IIa and Xa factors in the presence of AT-III was assessed for λ-CAR, λ-COS, λ-COS NP, and oligosaccharide of heparin (Hep OS) as an anticoagulant standard (Figure 6D,E). IC₅₀ were calculated from this data and it appears that λ-CAR inhibited to a significantly lesser degree than the Hep OS for both factors (Figure 6F), 0.14 mg·mL^–1 for Xa and 0.12 mg·mL^–1 for IIa, against 3 × 10^–4 and 9 × 10^–4 mg·mL^–1 respectively. However, it seems that the depolymerization eased the inhibition of the factor Xa for the λ-COS. Indeed, the IC₅₀ observed is 7.3 × 10^–2 mg·mL^–1, which is lower than that of the native form. However, the IC₅₀ for λ-COS NP was similar to that of the native form (0.19 mg·mL^–1), which might be due to steric hindrance preventing the specific sequence from binding to AT-III. For the inhibition of factor IIa, the reduction in chain length had a noticeable effect; the IC₅₀ for λ-COS was eight times higher than that of the native form, with an IC₅₀ of 0.8 mg·mL^–1. This result is consistent with the molecular weight of λ-COS, which was assessed at a M_n of 3.8 kDa, and was lower than that required to activate the binding of IIa. Concerning λ-COS NP, the inhibition of factor IIa did not exceed 25%, regardless of the concentration of OS, indicating that steric hindrance due to the structural arrangement of the oligosaccharides into NP reinforces the difficulty of binding and form between ATIII, IIa and OS.

Finally, cellular experiments were conducted to explore the potential toxicity of λ-COS NP on epithelial cells prior to in vivo experiments. For this, MTT cell viability (Figure 7A) and NP cellular uptake assays (Figure 7B) were performed in human embryonic kidney cells (HEK293) at various doses in the 0–200 μg·mL^–1 range and using two incubation times. HEK293 was selected as a convenient standard cell line, widely used in pharmacological assays, including preliminary evaluations of NP cytotoxicity.^72,73 Only λ-COS showed decreased metabolic activity after 24 h at 100 and 200 μg·mL^–1 concentrations which disappeared after 48 h, while no effect was observed for λ-COS NP at any incubation time. Observations of a decrease in metabolic activity in the early stages of incubation have been previously observed for free polysaccharides and OS, which can reach sublytic concentrations in HEK293 cells, initially affecting their viability, but not causing cell lysis over prolonged time.⁷⁴ In any case, this event was not observed for λ-COS NP, providing preliminary indications of the safety of this NP candidate. In addition, after a 24 h incubation with NP, HEK293 cells were fixed and stained with Prussian Blue after a 24 h incubation to reveal the presence of iron. As shown in Figure 7B, no internalization of λ-COS NP was observed.

Figure 7.

In cellular experiments: (A) Viability of HEK 293 measured by MTT after 24 and 48 h incubation, (B) NP uptake visualized by Prussian blue staining in HEK293 cells after 24 h incubation ([Fe] = 10 μg·mL^–1). Values represent mean ± SEM with N = 3 and n = 3, p values were calculated by one-way ANOVA with Tukey’s multiple comparison test.

3.7. MRI (1 T)-Based Biodistribution Studies in Mice and Imaging Applications Relating to Mn²⁺ Coupling Properties

In vivo MRI studies at 1 T were carried out in healthy mice (n = 9, Table S12) to assess the contrast performance of the λ-COS NP and their in vivo behavior, including vascular lifetime, biodistribution, and clearance from the body. In all cases, λ-COS NP were intravenously (i.v.) administered at a dose ranging from 1 to 3 mg·mL^–1 of (Fe + Mn), and the time course of the signals was recorded for up to 4 h (n = 6) or 24 h (n = 3). As shown in the T₁-weighted MRI images in Figure 8 and S13, several organs were brightened immediately after the i.v. injection of the NP—notably the heart, liver, and kidneys—thereby confirming the positive contrast capability of λ-COS NP. T₂-weighted gradient echo (FLASH) and T₁-weighted spin echo (MSME) sequences were also used, which led to similar results, although contrast enhancements were less pronounced (Figures S14 and S15).

Figure 8.

Biodistribution study in MRI at low field (1T) with a T₁-weighted gradient echo sequence of one representative case. Coronal views of upper abdominal region (A) and axial views of the heart and liver (B), additional coronal views of lower abdominal region (C), and axial views of the kidneys (D). i.v. administration of 100 μL λ-COS NP solution at [Fe + Mn] = 3 mg·mL^–1.

The intense signals found in the heart with the T₁-weighted sequences mainly accentuate the peripheral myocardium (Figures 8A,B, and S13), which is strongly indicative of a partial release of Mn²⁺ from the λ-COS NP once entering the blood circulation, consistent with previous in vitro experiments. Indeed, a peculiar application in the manganese-enhanced MRI (MEMRI) field relies on the avid uptake of Mn²⁺ ions by the voltage-gated calcium channels of the viable cardiomyocytes, where they can remain for hours.⁷⁵ This direct intracellular contrast, paired with the preferential uptake compared to stunned cells, has led to an increasing interest in the characterization and diagnosis of cardiac pathologies, offering an interesting alternative to gadolinium-based organic chelate contrast agents (GBCA), which allow only the assessment of the extracellular space.^76,77 Despite Mn(II) being an essential endogenous trace element for numerous biological functions and being rapidly cleared from the circulation by hepatocytes, an important issue for this kind of application remains in mitigating the toxicity from Mn overexposure, such as parkinsonism-like symptoms and cardiotoxicity.⁷⁸ For this, according to various articles,⁷⁹ the total Mn dose injected should first be limited to a concentration below those reported to cause detectable adverse effects that is estimated between 200 and 500 μM·kg^–1. The doses used in this study were approximately in the 25–95 μM·kg^–1 range (Table S12) and still, a significant contrast enhancement in various organs was achieved. Second, one common approach used to reduce the risks of Mn toxicity, while maintaining a suitably low level of free Mn²⁺ in circulation, is to complex the metallic ion within a ligand that allows controlled partial uncoupling in the plasma, facilitating its circulation as a protein-bound complex. In this way, a more persistent releasable Mn²⁺ source and buffered lower concentration for intracellular contrast can be achieved, with a predominant prolonged hepatobiliary clearance.^65,78,80 This approach even led to a marketed product based on manganese dipyridoxyl diphosphate (Mn-DPDP, Telascan), initially intended for use as contrast agents for liver and gastrointestinal imaging. It has since been superseded by the greater efficiency of GBCA and a lack of demand, but it is still applied in preclinical studies and new clinical trials, especially for heart diagnosis-based MEMRI.^65,77

Here, a comparable behavior occurred for λ-COS NP. Accordingly, in the heart, signals were maintained for up to 4 h, as shown in Figure 8A,B, due to the partial release of Mn²⁺ (limited to 20% in plasma as measured in vitro previously) and binding to blood proteins (as indicated in vitro previously by the increase of r₁ in serum), before uptake by the cardiomyocytes. Similar to Mn-DPDP and related to its initial purpose, a significant brightening in the liver also appeared immediately after i.v. administration (Figures 8A–C and S13), owing to Mn²⁺ uptake by hepatocytes, and lasted up to 4 h with a progressive decrease in the contrast enhancement ratio. In tandem, an intense signal was gradually detected in the gallbladder (Figures 8A,B and S13) and biliary tree, peaking at 2 h before vanishing entirely at 24 h. As a control and for the comparison of NP contrast performance, Multihance (gadobenic acid), a liver-specific contrast agent was used (Figure S16). As seen on the images, the MRI signal intensities were higher for the λ-COS NP than for the GBCA and lasted longer, even with a lower Mn dose administered (∼24 μM·kg^–1) compared to the Gd dose (∼116 μM·kg^–1). This observation is interesting, as GBCA are often preferred to Mn-based contrast agents due to their typical higher contrast efficiency achieved with a smaller dose of administered metallic ions, even if they do not offer the advantage of delayed contrast. Two hypotheses were proposed to explain this encouraging result: the partial excretion of Gadobenic acid through the urine, leading to a lower effective dose transiting in the liver; and/or an additional “contrast contribution” to the released Mn²⁺—originating positive contrast. This last might come, for instance, from the metabolized λ-COS NP encapsulating the extremely small ferrite cores, which can also produce a positive contrast. Finally, signals in the kidneys were also examined during the study (Figures 8C,D, and S13). Contrast enhancements were also visible until 4 h with a residual at 24 h, especially brightening the vascularised cortex/medulla region, whereas no detectable signal was found in the bladder throughout the time course of the experiments. Again, these results align well with previous MRI-based biodistribution studies of Mn-DPDP,⁸⁰ which revealed signals in this organ in the first hours after i.v. administration, attributed to free or complexed Mn²⁺ circulating in the blood, and also showed moderate accumulation (<20%) of the metabolized probe after 24 h.

3.8. MRI (7 T)-Based Biodistribution Studies in Mice for Studying the Influence of the λ-COS Coating on the NP In Vivo Behavior

The focus was then directed toward the λ-COS NP framework encapsulating the ferrite cores, to study the biointeractions and pharmacokinetic properties conferred to NP by this new carbohydrate-based coating. At low field, ferrite cores exhibit T₁-contrast ability due to their extremely small size; however, in our situation, such a contrast could therefore be confused with that of the free released Mn²⁺, making it difficult to discern the biodistribution of the intact λ-COS NP. In view of this, it was advisible to rely on the strong T₂ contrast properties found for the λ-COS NP at high field (r₂/r₁ ratio of 22.9 at 7 T), which mainly come from the ferrite cores, even if free Mn²⁺ should also exhibit minor T₂-relaxation shortening properties. In vivo experiments in mice with i.v. administration of λ-COS NP at three different doses ([Fe + Mn] at 0.8, 1.3, and 1.5 mg mL^–1, n = 2 for each, n = 6 total) were therefore repeated using a 7 T scanner, and images were recorded by a MSME (Multi slice Spin echo) sequence to build the T2 relaxometric maps before injection, at postinjection (0.5–2 h), and at 24 h. See Table S12 for details about doses and times of acquisitions. Representative T₂ maps obtained in mice are shown in Figures 9A and S17, and the mean T₂-values of selected ROIs in different organs were calculated for each dose administered, pooling together all the precontrast and postcontrast points of the two mice (Figures 9B and S18). Regarding the liver, a strong decrease in the T₂ values was observed post injection (0.5–2 h) in all cases, which only partially recovered after 24 h. It has been previously shown in the 1-T experiments that released Mn²⁺ were already eliminated after 24 h in the intestinal tree and should not, therefore, have been contributing to the MRI signals anymore. It is for this reason that this T₂-reduction that was still seen at that time is particularly interesting, as it strongly indicates that a part of intact λ-COS NP was taken up by hepatic macrophages. With respect to the change in the T₂-values in the kidneys, a significant decrease was also observed postinjection (0.5–2 h). This is more likely due to circulating NP, detected because of the high vascularisation of the cortex and medulla,⁸¹ rather than renal clearance of λ-COS NP through the urinary system. Indeed, the HD of NP was larger than the renal filtration threshold (estimated at approximately 8 nm), and no T₂ shortening was observed in the bladder, making this possibility improbable. Interestingly, T₂-shortening in the cortex and medulla was detected, even 2 h after the lowest dose was administered (Figure S18), which suggests that λ-COS NP have blood pool properties that are potentially advantageous for targeting pathological tissue. However, this hypothesis should be taken with caution because released Mn²⁺ could also make a small contribution to the signal and be confused with that of the intact λ-COS NP. This could explain the T₂ values found in the kidneys at 24 h, which had almost recovered to their basal values, but with a small reminiscent signal that may come from metabolized forms of Mn²⁺, consistent with the previous 1 T MRI experiments. Finally, to round out the picture, it can be noticed that a small decrease in T₂ was detected in the spleen, another organ commonly involved in NP elimination.⁸² It can be reasonably attributed to only λ-COS NP, as released Mn²⁺ is not supposed to accumulate in this organ,⁸⁰ and, besides, it was not detected in 1 T MRI.

Figure 9.

Biodistribution study in MRI at high field (7T) with (A) T₂-Relaxometric maps in cross sections obtained from a MSME sequence (TE varied in 20 steps from 8 to 160 ms, TR = 4000 ms, FA 90°). (B) T₂-value quantification in selected ROI of different organs, pooling together the contrast points of two mice (i.v. administration of 70 μL λ-COS NP solution: [Fe + Mn] = 1.5 mg mL^–1).

3.9. Ex-Vivo Analyses

To further explore the biodistribution of λ-COS NP and their metabolites (especially released Mn²⁺), ex vivo analyses were conducted. For this, the main organs of the RES system were collected and perfused at 3 h (n = 2) and 24 h (n = 1) following i.v. administration of NP (see Table S12 for doses). A histological analysis of tissue sections, stained with hematoxylin–eosin, was first performed as a preliminary indication of the potential toxicity induced by λ-COS NP or free Mn²⁺. No obvious pathological changes in the main organs were noticed in the images, either at 3 h or at 24 h (Figures 10A and S19, respectively). Next, Prussian blue staining was performed to reveal the in vivo iron-containing λ-COS NP distribution. At 3 and 24 h, numerous distinguishable blue dots were detected in the liver tissue corresponding to the Fe contained in the ferrite cores and colocalizing with the Kupffer cells (Figures 10A and S19). This confirms the previous results obtained with the MSME-based 7 T MRI, which showed that a part of the λ-COS NP was eliminated in the hepatic macrophages. Iron-containing cores were also detected in the lungs at 3 h, and to a minor extent at 24 h, indicating that NP lodged temporarily in small lung capillaries—a phenomenon frequently observed for NP with a HD up to 50 nm. Unfortunately, it was impossible to visualize any accumulation of NP in the spleen, as it already contained many endogenous iron forms that had been detected on the control image. In the other organs examined, no blue dots were clearly noticeable in either the intestines at 3 h, which may indicate that the “intact” NP did not undergo hepatobiliary clearance such as Mn²⁺, or in the heart, confirming that only released Mn²⁺ was taken up by myocytes. Regarding the kidneys, the absence of blue dots at 3 and 24 h confirmed that 7 T-MRI signals detected were more likely due to circulating NP (flushed during the ex vivo-perfusion step) than to accumulated NP in the tissue.

Figure 10.

(A) Histopathology in main organs 3 h after injection, stained by hematoxylin and eosin and Prussian blue, observed by optical microscopy at 20× and 40×. Injected volume: 70 μL, [Fe + Mn] = 1.5 mg·mL^–1 (n = 2). Black and white arrows point out the blue stains on Prussian blue images. (B) Mn and (C) Fe contents in the organs, measured by ICP–MS and expressed in mass (μg) of Mn per g of dry tissue. The data are mean ± SEM of n = 2 mice at 3 h and n = 4 mice at 24 h.

Mn and Fe were quantified in the organs by ICP–MS to provide information about the biodistributions of the Mn released and λ-COS NP’ associated ferrite cores, respectively. Concerning Mn dosage (Figure 10B), average Mn concentrations (μg/g of dry tissue) were higher in the kidneys and liver 3 h after the injection (n = 2) compared than at the physiological basal level. This result is in good agreement with the pharmacokinetic properties observed in the 1 T MRI-based biodistribution studies, where released Mn²⁺ from the NP was still circulating at that time, combined with avid uptake by hepatocytes. A decrease in the Mn level, although still above the basal level, was then obtained at 24 h in both organs (n = 4). Again, this fits well with the 1 T-MRI signals found at that time in the kidneys. It indicates the weak accumulation of Mn in this organ, while the higher concentration in the liver may come from either remaining Mn²⁺ ions, or from the λ-COS NP ferrite cores found in hepatic macrophages that also contain a small amount of Mn doping. Surprisingly, no significant enhancement in Mn concentration was found at 3 h in the heart to align with their uptake by the voltage-gated calcium channels of the cardiomyocytes observed on 1 T MRI. Concerning iron dosage (Figure 10C), higher average Fe concentrations compared to basal values were found in the spleen and liver at both 3 h (n = 2) and 24 h (n = 4). It should be noted here that the concentration is expressed in μg/g of dry tissue and should be balanced in accordance with the organ weights (much higher in the case of the liver). This is consistent with the presence of λ-COS NP encapsulating ferrite cores in the Kupffer cells, as well as their accumulation in the spleen, which was previously observed during 7 T MRI-based biodistribution experiments and confirmed by histology. In Kidneys, no additional iron level compared to basal was noticed at either time, confirming that 7 T-MRI signals detected at 3 h, in particular, were more likely due to circulating NP (flushed at the ex vivo-perfusion step) than to NP accumulated in tissue. Finally, a transient excess of iron was measured in the lungs 3 h following the NP injection, before a decrease was observed toward the baseline value by 24 h that was consistent with the histological observations.

To complete the ex vivo analysis, ICP–MS was also carried out on urine samples collected during the different experiments and did not reveal any significant change in Mn or Fe concentration. This provides an additional indication regarding the absence of renal clearance for this probe and its associated metabolites (data not shown). Measurements were also done on the intestine and collected faeces, but results should be interpreted with caution as mice were not fasted in this study. While Mn concentration (μg/g of dry tissue) remained roughly similar in the intestine (Figure S20), some faecal samples exhibited unusually high levels of Mn, in accordance with the clearance of released Mn²⁺ through the intestinal tract (data not shown). Uncovering information from the Fe dosage was more challenging. This was due to the already high basal level in faeces, even if it appears that Fe levels were surprisingly lower (data not shown), while concentration in the intestine displayed an increase in concentration (Figure S20). It could be suggested that degradation of ferrite cores confined in hepatic macrophages could initiate a progressive release of metabolized iron retained in the intestine.

4. Conclusions

In this study, depolymerised derivatives of λ-CAR were evaluated as potential NP scaffolds for exploring novel chemistries, specific functionalities, and in vivo biointeractions. Initial results in cell-based models and in vitro assays indicated that λ-COS and λ-COS-NP exhibited substantially higher biocompatibility than native λ-CAR, particularly demonstrating reduced pro-inflammatory activity. The use of λ-COS also improved solubility compared to λ-CAR and decreased viscosity, thereby simplifying NP synthesis. Under microwave-assisted conditions, the resulting nanostructure was a λ-COS-based hybrid NP, in which the λ-COS scaffold served both as a coating to encapsulate ferrite cores and as a macromolecular chelator for Mn²⁺ ions. Advanced electronic microscopy and spectroscopy techniques revealed that the cores formed crystalline structures in a mixed spinel arrangement. In parallel, substitution with Zn²⁺ verified the specific stabilization of Mn²⁺ ions by chemical groups within the organic matrix rather than by nonspecific adsorption at the surface of the NP. Subsequent magnetometry studies confirmed the presence of both ferrite cores and a population of complexed Mn ions in the NP. Relaxometric measurements further showed that the NP exhibited different contrast properties in MRI, depending on the applied magnetic field and the contributions from free Mn²⁺, complexed Mn²⁺ and ferrite cores. Finally, in vitro stability tests demonstrated that the complete NP remained stable in aqueous media, while approximately 20% of the Mn²⁺ payload was released in serum.

In vivo MRI studies in mice using a 1 T scanner indicated that the λ-COS NP, in their “Mn-releasable complex” configuration, are suitable for a typical application of Mn²⁺-based intracellular contrast in MEMRI, specifically for imaging the myocardium and the hepatobiliary system. The observed biodistribution and clearance profiles were similar to those typically reported for Mn-DPDP, an FDA-approved ligand that controls the release of Mn²⁺ in vivo, thereby reducing potential toxicity concerns. Further MRI examinations at 7 T along with ex vivo studies suggest that the λ-COS NP, in their “ferrite core coating” configuration, exhibit a good vascular lifetime relative to other polysaccharide-based NP and undergo at least partially, clearance by hepatic macrophages.

Future investigations are necessary to more clearly distinguish these two functionalities. For instance, synthetizing λ-COS-based NP without ferrite cores will allow a deeper exploration of the metalation chemistry, particularly as macromolecular Mn²⁺ chelates. Such a study could confirm the gradual release of Mn²⁺ properties in serum to avoid overexposure, clarify the role of submetabolites following i.v. administration (including the possibility that retained Mn²⁺ in λ-COS NP may lead to distinct biodistribution in pathological tissue), and assess the overall contrast performance as an alternative to GBCA. Additionally, different synthetic routes could be employed to prepare “single” λ-COS coated metallic NP cores. This approach would aid in clarifying how the oligosaccharide coating influences pharmacokinetic properties, biodistribution, and clearance of metallic NP, while also enabling further evaluation of potential biofunctional roles of λ-COS in NP formulations, such as targeted delivery or therapeutic activity.

Nevertheless, the λ-COS NP exhibiting this “dual feature” may offer potential advantages for specific bioapplications. For example, the probe could function as a theragnostic agent responsive to the tumor microenvironment, releasing Mn²⁺ ions more efficiently under acidic conditions. The resulting strong contrast enhancement may confirm accumulation at the tumor site, while additional bioactivities could arise from the λ-COS coating, the ferrite core, or the released Mn²⁺.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.biomac.4c01747.

• Figure S1: SEC-HPLC spectra of λ-COS, Figure S2: FT-IR and UV–visible spectra of metals- λ-COS coordination complex, Figure S3: Mn released assessment, Figure S4: TEM and HRTEM images, Figure S5: HRTEM images associated with EDS in maps, Figure S6: HRTEM images associated with EDS in line, Figure S7: diffraction pattern and d-spacing, Table S8: estimation of number of COS Fe and Mn per NP, Figure S9: XPS characterization, Figure S10: magnetic moment at different temperatures, Figure S11: phantoms at 1T and 7T, Table S12: table listing the in vivo experiments, Figure S13: MRI (1T) images (T1-GE sequence), Figure S14: MRI (1T) images (T2-GE sequence), Figure S15: MRI (1T) images (MSME sequence), Figure S16: Comparison of contrast performance with multihance, Figure S17: MRI (7T) images, Figure S18: T₂-value quantification in selected ROI of different organs, Figure S19: histopathology, 24 h after the injection, Figure S20: Mn and Fe content in intestine measured by ICP–MS (PDF)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.