Polydopamine-Functionalized Magnetoliposomes as Magnetic Probes for Lectin–Glycan Interaction Analysis through Brownian Relaxation

Abstract

Polydopamine-coated magnetoliposomes (PDA-MLs) are adhesive biocomposite nanomaterials commonly employed in intracellular applications such as controlled drug delivery, theranostics, and biosensing. However, the influence of their adhesive properties on lectin–glycan interactions remains unexplored. In this study, we developed PDA-MLs as magnetic sensing labels for the rapid detection of lectin–glycan binding using alternating current (AC) magnetization measurements in liquid media. The adhesive PDA-ML surface enables facile biomolecule adsorption, facilitating their integration into biosensors. The effects of dopamine-polymerization time on the morphological and functional properties of PDA-MLs were systematically investigated. Extended polymerization resulted in a thicker PDA layer that enhanced lectin immobilization on the ML surface. Multivalent N-glycan-clustered albumin facilitated cross-linking among lectin-immobilized PDA-MLs, leading to remarkable changes in the AC magnetization, compared with interactions with monovalent Gal-N-glycan and Sia-N-glycan. Specific lectin–glycan pairs produced distinct changes in AC magnetization spectra, whereas mismatched combinations showed minimal changes. These findings reveal that structure–property relationships are critical to the design of PDA-ML-based biorecognition systems, highlighting their potential in biosensing and bioanalytical applications.

Introduction

The detection of biomolecules, such as proteins, nucleic acids, lipids, and glycans, has become increasingly important for the rapid and early detection of disease biomarkers, enabling infection control and prompt clinical treatment. Current detection methods based on electrochemistry, optics, , fluorescence, , and magnetics have made significant advances in biosensing. In this context, magnetic nanoparticle (MNP)-based biosensing has emerged as a promising solution for detecting and quantifying circulating biomarkers in bodily fluids. This technique offers the significant advantage of selective biomolecule identification with minimal background noise. Most biological substances are nonmagnetic; hence, MNP-based biosensing significantly minimizes interference from nontarget substances, enabling more precise and reliable analyses. −

MNPs have been widely employed as magnetic sensing labels for the detection of biological targets through their functionalization with specific probe molecules such as proteins and antibodies. The interaction between probe-conjugated MNPs and target molecules can be magnetically detected using magnetic-particle spectroscopy (MPS) and alternating current (AC) magnetization measurements. , The principle underlying these measurements is the physical rotation of superparamagnetic MNPs dispersed in a fluid under an AC magnetic field, a phenomenon known as Brownian relaxation. The resulting hydrodynamic and magnetization responses of MNPs provide precise information about their hydrodynamic size and the local viscosity of the surrounding medium. − In Brownian relaxation-based biosensing applications, interactions between MNPs and biomolecules modulate the particles’ hydrodynamic volume and surrounding viscosity, resulting in detectable changes in frequency-dependent magnetization under AC magnetic fields.

Beyond biomolecular interactions, changes in the AC magnetization spectrum, such as peak shifts and broadening, indicate changes in the hydrodynamic behavior of MNPs due to aggregation in biological environments containing various salts. − In such conditions, diminished electrostatic repulsion can compromise dispersion stability and mobility, thereby resulting in particle aggregation or sedimentation. , Consequently, enhanced control over colloidal stability in suspensions is urgently required when using MNPs for Brownian relaxation-based biosensing. In this context, lipid-based structures have emerged as particularly promising candidates. Liposomes, biocompatible entities composed of phospholipid bilayers suspended in an aqueous medium, have been investigated as potential nanocarriers that can enhance the biocompatibility and colloidal stability of MNPs.

Significant advancements have been achieved in the development of lipid-based nanoassemblies created by conjugating MNPs with liposomes, known as magnetoliposomes (MLs). The surrounding lipid bilayer enhances systemic biocompatibility and protects the MNPs from aggregation and oxidation while preserving their magnetic properties. , The unique magnetic characteristics of MLs facilitate straightforward manipulation and separation upon exposure to external magnetic fields. Furthermore, they possess inherent advantages, including high stability and the capacity for functionalization with specific ligands, rendering them exceptional candidates for diagnostics, imaging, , drug delivery, , and therapeutic applications. , By adjusting the lipid composition of the vesicles, adsorption and interaction with biological molecules can be further enhanced.

Thus, these hybrid nanostructures effectively combine the beneficial characteristics of both MNPs and lipid carriers, making them appealing platforms for analyzing biological molecules. Alonso et al. demonstrated the use of MLs in magnetic biodetection based on magnetoreactance measurements. Most previous studies utilizing MLs for biosensing have primarily focused on immunomagnetic separation and enrichment methods. , However, little progress has been made in advancing techniques that use MLs as sensing labels. A key challenge is the effective incorporation of MNPs into liposomes. We previously demonstrated the potential of MLs as magnetic sensing labels by analyzing their AC magnetization. The potential of MLs as magnetic labels can be further realized by tailoring their surfaces, which serve as an interface with the biological environment. The surface characteristics of nanostructured materials ultimately determine their stability and behavior in vivo. Polydopamine (PDA) is an adhesive biomaterial coating produced via the in situ self-polymerization of dopamine (DA) under mild alkaline conditions. The adhesive surface promotes facile biomolecule adsorption and effective intracellular delivery for biomedical applications such as receptor targeting and biosensing. − Lectin–glycan interactions play pivotal roles in diverse biological processes, including cellular recognition, adhesion, signal transduction, immunomodulation, inflammation, and host–pathogen recognition. However, investigating these interactions presents significant challenges due to their noncovalent and relatively weak nature, typically characterized by dissociation constants, K _D, in the mM range. An effective strategy to enhance binding affinity involves multivalency, which substantially improves the weak affinity of individual monovalent receptors for their corresponding glycans through the simultaneous engagement of multiple binding sites. , Existing methods for detecting lectin–glycan interactions include quartz crystal microbalance (QCM), microarrays, surface plasmon resonance (SPR), and isothermal titration calorimetry (ITC). These techniques are effective for analyzing lectin–glycan interactions; however, limited progress has been made in developing techniques for assessing weak interactions underlaying natural binding events in whole colloidal systems. This is particularly important because existing methods rely on immobilizing receptors on solid supports, which can hinder natural binding behavior and require large sample volumes. To overcome these challenges, we propose a versatile technique based on magnetic measurements of the Brownian relaxation behavior of PDA-coated MLs.

Given the exceptional adhesiveness, biocompatibility, stability, facile synthesis, and functionalization of PDA, we hypothesize that appropriately formulated PDA-coated ML surfaces can serve as alternative linkers to enable lectin accessibility, thereby facilitating multivalent glycan binding and enhancing the investigation of lectin–glycan interactions. In this study, we investigated the dynamic magnetic behavior of PDA-MLs upon the introduction of lectin–glycan (receptor–analyte) moieties to form glycan clusters with MLs (Figure ). PDA functions as a linker, enhancing protein immobilization through its robust adhesive properties. Prior to PDA incorporation, we evaluated the colloidal stability and morphology of the MNPs and MLs. Subsequently, we synthesized a series of PDA-MLs by integrating DA into the ML surfaces (Figure A). We employed glycan-modified albumin (N-glycoalbumin) to demonstrate biomolecule detection, specifically lectin–glycan interactions. N-glycoalbumin, comprising human serum albumin (HSA) conjugated with asparagine-linked glycans, was utilized to mimic natural glycan patterns and provide multivalent bindings (Figure B). We examined the formation of aggregates between PDA-MLs and glycans at varying lectin concentrations through AC-magnetization spectral analysis (Figure C), demonstrating the potential of our proposed magnetic sensing platform for biosensing applications. This platform was developed for analyzing glycan patterns on proteins in colloidal systems, and thus, lectins immobilized on ML surfaces are treated as receptors. As shown in Figure D, we employed two types of N-glycoalbumin: α­(2,6)-sialylated (Sia-N-glycoalbumin) and galactosylated (Gal-N-glycoalbumin), along with their corresponding monoglycan counterparts, including sialylated N-glycan (Sia-N-glycan) and galactosylated N-glycan (Gal-N-glycan). This approach enables differentiation between multivalent and monovalent lectin–glycan binding events based on their aggregation states. Notably, Sia-N-glycoalbumin has nine sugar chains, whereas Gal-N-glycoalbumin has eight sugar chains.

1.

Schematic representations of lectin–glycan (receptor–analyte) interactions between polydopamine-coated magnetoliposomes (PDA-MLs) and N-glycoalbumins. (A) Formation of PDA-ML–N-glycoalbumin clusters involves: (i) synthesis of MLs, (ii) PDA layer formation on ML surface, (iii) lectin modification on PDA layer, and (iv) conjugation of N-glycoalbumins to lectin-immobilized PDA-MLs. (B) MLs and N-glycoalbumins conjugate to form clusters that are cross-linked via multivalent bindings. (C) Frequency spectra of AC magnetization for supernatant are measured to estimate the amount of ML–N-glycoalbumin clusters in the supernatant and the cluster sizes by real part M′ and imaginary part M′′ of magnetization, respectively. (D) Two types of artificial N-glycoalbumins are prepared by conjugating monoglycanssialylated N-glycans and galactosylated N-glycansto human serum albumins (HSA): (left) Binding of Sambucus nigra agglutinin (SNA) lectins to α­(2,6)-sialylated N-glycoalbumin, and (right) binding of peanut agglutinin (PNA) lectins to galactosylated N-glycoalbumin. This figure was created with BioRender.com.

Materials and Methods

Synthesis of Magnetic Nanoparticles

Iron-oxide nanoparticles (IONPs) were synthesized using the solvothermal method. Briefly, 0.7 g of iron­(III) acetylacetonate (14024-18-1, Sigma-Aldrich, USA) was dissolved in 20 mL of tetraethylene glycol (112-60-7, Sigma-Aldrich, USA) at 80 °C. The solution was transferred to two 10 mL stainless pressure vessels and heated to 250 °C for 15 h in an electric furnace. After the reaction, the synthesized IONPs were washed once with acetone, collected by centrifugation at 20700g, and rinsed twice with a mixture of deionized (DI) water and ethanol (1:3 v/v%). The synthesized IONPs were dispersed in an aqueous solution containing 1.0 g of trisodium citrate (197-06025, FUJIFILM Wako Pure Chemical Corp., Japan) in 60 mL of DI water. This dispersion was achieved by using a homogenizer at 70 °C; the IONPs were modified with citric acid (CA), resulting in a negatively charged coating. After the surface-modification process, the CA-modified IONPs were washed three times with a mixture of DI water and ethanol (1:3 v/v%) and dispersed in 5 mL of DI water. The CA-modified IONPs were transferred to water by exchanging surface CA with (3-aminopropyl) trimethoxysilane (APS) to obtain positively charged IONPs, as reported previously.

Production of Magnetoliposomes

Magnetoliposomes were prepared using the double-emulsion method. , Aliquots of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC; Avanti Polar Lipid, USA), 1,2-dis-tearoyl-sn-glycero-3-phosphocholine (DSPC; Avanti Polar Lipid, USA), cholesterol (Sigma-Aldrich, USA), and N-(methyl­polyoxy­ethylene oxycarbonyl)-1,2 distearoyl-sn-glycero-3-phosphoethanolamine, sodium salt (DSPE-PEG 2000; SUNBRIGHT® DSPE-020CN, NOF Corp., Japan) supplied as chloroform solution were prepared as a stock solution at 10-mM lipid concentrations. The lipid-to-MNP ratio was set at 2:1 (mol:mol). In brief, the MNPs were dissolved in an aqueous solution and then transferred into a mixed solution of amphiphilic lipids at a molar ratio of 13:6:7:0.8 (DPPC:cholesterol:DSPC:DSPE-PEG). An ultrasonic homogenizer (UX 300, MitsuiElectric Co., Japan) was then used to create an oil-in-water (w/o) emulsion, encapsulating the MNPs within the inner water droplets. Subsequently, an excess of HEPES buffer (10 mM HEPES, 150 mM NaCl, pH 7.4) was added as the external aqueous phase, producing a double-emulsion system characterized by a two-phase solution (w/o/w) (Figure S1). The lipids formed around the water droplets, effectively entrapping the MNPs within the liposomes. Finally, rapid evaporation of the organic solvent resulted in the formation of a double micelle structure of magnetoliposomes, yielding a final one-phase solution.

The suspensions were extruded through a polycarbonate filter (Nucleopore track-etched hydrophilic membranes, Cytiva, MA, USA) with a decreasing pore diameter of 0.4 and 0.2 μm (five passages through each). The unencapsulated particles were removed using a removable magnetic bar (PureProteome magnetic stand, Merck Millipore, Germany) and centrifuged at 4000g for 5 min. The purified ML solution was collected and stored at 4 °C until further modification.

Functionalization of Magnetoliposome Surface with Polydopamine

PDA-MLs were prepared by adding 0.2 mg/mL of DA solution in Tris buffer (pH 8.5) into the ML suspension. This mixture was stirred under temperature-controlled conditions for 1, 4, and 24 h. After the desired incubation time, the sample was purified to remove unreacted DA monomers and any molecular intermediates from the polymerization reaction. This was achieved through dialysis against DI water using a cellulose nitrate membrane with a molecular weight cutoff (MWCO) of 6–8 kDa (Pur-A-Lyzer Maxi, Merck, Germany) for 24 h, followed by centrifugation at 4000g for 15 min. The purified PDA-MLs were stored at 4 °C until characterization, as PDA-MLs remain stable for at least 2 weeks when refrigerated.

Materials Characterization

Transmission electron microscopy (TEM; H-8100 Hitachi High-Tech Corp., Japan) at an accelerating voltage of 200 kV was used to determine the core size and shape of IONPs. For MLs, a negative staining method was employed. The PDA layer and negative staining solution can impart a dark color to the edges of the liposome surface. The crystallinity of the iron oxide was determined using X-ray diffraction (XRD; RINT-2000, Rigaku, Japan), employing Cu Kα radiation (λ = 1.54 Å).

Fourier-transform infrared (FT-IR; FT-IR-6600, JASCO Corp., Japan) spectra in attenuated total reflectance (ATR) mode were obtained for APS-IONPs, CA-IONPs, DA, liposomes, MLs, and PDA-MLs in powder form after freeze-drying. An ultraviolet–visible (UV–vis) spectrophotometer (UV-670, JASCO Corp., Japan) was used to measure the absorption spectrum. After each polymerization interval, the DA solutions and PDA-MLs were evaluated under identical dilution conditions.

The hydrodynamic sizes of the synthesized IONPs and MLs were measured in either water or HEPES buffer using dynamic light scattering (DLS; SZ-100 V2, HORIBA Ltd., Japan). The reported hydrodynamic sizes are the “Z-average” values from three independent measurements. The refractive indices used for IONPs and liposomes were 2.42 and 1.47, respectively. The zeta potentials of both MNPs and MLs were measured using the same DLS equipment. These values were determined as the averages of three independent measurements.

The iron content in MLs was measured by using inductively coupled plasma-atomic emission spectrometry (ICP-AES; ICPS-8100, Shimadzu Corp., Japan). ML samples were dispersed in 1 mL of HNO₃ (69%) and heated in an oil bath at 80 °C for 1 h. After digestion, the samples were diluted with DI water to achieve a final volume of 10 mL prior to the measurements.

The frequency dependence of AC magnetization was measured using a physical property measurement system (Quantum Design Corp., USA) at a magnetic-field strength of 0.8 kA/m. Measurements were conducted at 300 K across a frequency range of 1–10 kHz using suspension specimens with a volume of 100 μL. Spectra of AC magnetization were recorded as a function of frequency, where the AC magnetization (emu/cc; magnetic moment per unit volume of the suspension), M = M′ – iM′′, comprises the real part M′ corresponding to the in-phase rotation of MLs with respect to the AC magnetic field and the imaginary part M′′ corresponding to the out-of-phase component, accounting for the dissipation of energy as heat.

Analysis of Lectin–Glycan Interactions Using AC Magnetization Measurements

Lectin–glycan interaction analyses were performed using lectin-immobilized PDA-MLs (receptor on sensing surface) with glycans or N-glycoalbumins (analyte) based on AC magnetization measurements. A stock solution of lectin (Vector Laboratories, Inc., CA, USA) was prepared at 1.0 mg/mL for both peanut agglutinin (PNA) and Sambucus nigra agglutinin (SNA) in DI water. Various lectin concentrations ranging from 0.00 to 50.0 μg/mL were added to the PDA-ML solutions, and the mixtures were incubated at 4 °C for 2 h to immobilize the lectin onto PDA-MLs. The mixture was centrifuged at 4000g for 15 min to remove the unbound lectin. The resulting precipitate was redispersed in fresh HEPES buffer (10 mM HEPES, 150 mM NaCl, and 1 mM CaCl₂). Glycan stock solutions (Glytech, Inc., Japan) were prepared at 10 μM by using Sia-N-glycan, Gal-N-glycan, Sia-N-glycoalbumin, and Gal-N-glycoalbumin in DI water. The N-glycoalbumins used in this study were prepared according to the previously reported protocol. [Ref ] The lectin-immobilized PDA-ML suspension was mixed with a 1 μM glycan solution at equal volumes. This mixture was incubated for 30 min at 300 K, after which the AC magnetization was measured. The frequency spectra of MLs are modulated by interactions with glycans or N-glycoalbumins. Clustering or aggregation resulting from these interactions induces changes in hydrodynamic size, leading to spectral modulation. Control experiments were conducted to assess the specificity of lectin–glycan interactions, including combinations of nonimmobilized PDA-MLs and glycans and mismatched lectin–glycan combinations such as PNA with Sia-N-glycan or Sia-N-glycoalbumin and SNA with Gal-N-glycan or Gal-N-glycoalbumin.

Results and Discussion

Synthesis and Characterization of IONPs and MLs

The primary size of IONPs was determined to be 14.7 ± 0.2 nm using TEM (Figures S2A,B). The XRD pattern indicates that the APS-IONPs exhibit typical spinel-crystal structures, as evidenced by the peak positions (Figure S2C). The crystallite size calculated from the XRD result was 14.6 ± 1.5 nm.

The FT-IR spectra of the CA-IONPs display characteristic bands with O–H stretching above 3100 cm^–1, C–H vibration at 3000 cm^–1, and a range of 1000–1700 cm^–1 associated with coating signatures, including CO for carboxylic acid on CA, and Si–O in the APS structure. Bands below 1000 cm^–1 are attributed to the Fe–O stretching modes of iron oxide (Figure S2D). The peak at approximately 1095 cm^–1 corresponds to Si–O–Si stretching, and the attenuation of the broad O–H peak indicates that APS-IONPs have been successfully modified on the particle surface. The AC magnetization of APS-IONPs showed a peak frequency above 10 kHz (Figures S2E,F). It should be noted that Néel relaxation, the rotation of the magnetic moment inside MNP, is negligible in all the measurements because frequency spectra of AC magnetization obtained from completely dried MNPs (Brownian relaxation is blocked) revealed that real and imaginary parts are flat, nearly zero in the range of 10–10 kHz, suggesting that Néel relaxation does not contribute within this frequency range (Figure S3).

The TEM images indicated a size of 162.4 nm for the MLs and confirmed that APS-IONPs were encapsulated within the liposomes (Figures S4A,B). DLS analyses revealed that MLs present in the supernatant after centrifugation had a hydrodynamic diameter (Z-average) of 173.3 ± 5.3 nm. An investigation of the supernatant collection, instead of the precipitate, indicated that the resuspended precipitate exhibited a larger average hydrodynamic diameter than that of the uncentrifuged suspension, ranging from approximately 500 to 1000 nm. This observation is likely attributed to smaller liposomes remaining in the supernatant during centrifugation, suggesting that aggregation or disruption phenomena occur as a result of the increased shear forces associated with high rotational speeds. The blank LPs displayed a hydrodynamic size of 99.3 nm ± 1.7 nm. The nanoparticles coated with either CA or APS exhibited hydrodynamic sizes of 22.5 ± 1.6 nm and 31.8 ± 1.1 nm, respectively (Figure S4C). The zeta potentials of the ML, LP, APS-IONP, and CA-IONP were measured as +16.1 ± 0.6, +19.3 ± 0.3, +26.0 ± 1.3, and −36.6 ± 0.7 mV, respectively (Figure S4D). A quantitative chemical analysis of Fe using ICP-AES revealed that the Fe concentration in the ML suspension was approximately 0.10 mg/mL.

PDA-MLs were prepared using a modified two-step procedure based on previously developed protocols. Small unilamellar MLs obtained using the method described in the previous section were incubated in a slightly alkaline DA solution at pH 8.5 for three different polymerization durations. After each polymerization period, the PDA-MLs were purified by dialysis to remove excess salt, followed by centrifugation to eliminate residual DA. A low concentration of DA solution (0.2 mg/mL) was employed in this study to minimize the self-polymerization of DA and control the thickness of the PDA shell. By maintaining this suspension under constant and mild stirring, the MLs were coated with PDA through the oxidation of DA. According to the reported structural model of PDA, the polymerization of DA predominantly converts amines to indoles and diphenols to quinones. In the UV–vis absorption spectrum of the DA solution, the characteristic absorption peak at a wavelength of 280 nm gradually diminished. Concurrently, after 24 h of incubation, the peak at 280 nm shifted to longer wavelengths (300–350 nm) or broadened due to increased conjugation within the aromatic and quinoid structures in PDA. A new broad absorption band emerged in the 400–500 nm range (Figure A). Additionally, the scattering contribution in this band increased, indicating a progressive growth in particle size with longer polymerization times (Figure B).

2.

UV–vis absorption spectra of (A) DA and (B) ML and PDA-ML at various polymerization times; (C) ATR-FTIR spectra of DA, PDA-ML, ML, and LP. Reaction conditions: DA = 0.2 mg/mL; T = 25 °C. Numbers represent wavenumbers of the absorption peak position in the spectra.

FT-IR spectra for DA, PDA-ML, ML, and LP all exhibited strong absorption in the 3500–2850 cm^–1 range. This range is characteristic of the O–H stretching vibration (Figure C). , In the PDA-ML spectrum, the characteristic peaks of monomeric DA between 750 and 900 cm^–1 (C–H bending of aromatic compounds) disappeared, whereas a band appeared at 3346 cm^–1, indicating the stretching vibrations of phenolic aromatic O–H and N–H. Additional absorbance bands, located at 3182 cm^–1 (C–H stretching of aromatic compounds) and 1035 cm^–1 (C–O stretching of phenolic group), emerged after modification with PDA. The stretching vibration of the aromatic ring and bending vibration of N–H were observed at a peak of 1605 cm^–1, whereas the stretching vibration of phenolic C–O was at 1510 cm^–1. These observations confirmed the successful deposition of PDA in accordance with previous literature , (Figure C, blue trace). Furthermore, the FTIR spectra of the initial LP and ML (Figure C, red and black traces) clearly highlighted the CO ester stretching band of the phospholipids at 1730 cm^–1, along with bands at 2920–2850 cm^–1 (C–H stretching) and 1240 cm^–1 (phosphate group), which are characteristic of the liposome spectra.

The hydrodynamic size and zeta potential of the PDA-MLs increased with an increasing polymerization time (Table ). Following the PDA coating, the zeta-potential value of the vesicles remained slightly negative, which is consistent with previously reported data for PDA nanoparticles suspended in nearly neutral solutions. , The isoelectric point of PDA is estimated to be in the pH range of 3–4. At pH values below 4, the amino groups in the polymeric structure become protonated, leading to a positive surface charge. Conversely, the phenolic groups are deprotonated at pH values above the isoelectric point, resulting in negatively charged PDA. As the average diameter of PDA-MLs increased, their size distributions remained unimodal. This observation suggests that DA polymerization occurred around the liposome surface, forming PDA-ML structures rather than discrete PDA particles. When PDA is generated using oxidants such as O₂, a thin film of PDA is deposited on the walls of the reaction tubes and on the target substrate. In addition, insoluble black precipitates, termed PDA particles, form and settle at the bottom of the reaction vessel. These PDA nanoparticles can also be produced without a templating agent by directly oxidizing catecholamines. The presence and concentration of a templating agent strongly influence the amount of PDA particles produced. Notably, lipid vesicles exert a pronounced templating effect on the PDA formation. In these cases, large PDA aggregatestypically observed in the absence of liposomeswere not observed nor was there segregation of PDA nanoparticles. Furthermore, experimental observations showed that using liposomes as a surface template reduced the deposition of the PDA layer on the reaction tube and no PDA particle precipitation occurred at the bottom of the tube. The results also indicated that the polymerization time plays a crucial role in determining the colloidal stability and morphology of PDA-MLs. TEM images revealed that the prepared MLs were spherical and exhibited a darker surface layer, which indicates the successful deposition of PDA on the liposome surface (Figure ). As the polymerization time increased, the thickness of the PDA layer also increased. At lower magnifications, the images depict well-dispersed PDA-MLs (Figure A,C). However, only a few vesicles were observed in the PDA-ML samples after polymerization for 24 h (Figure E). The image of the PDA-ML sample polymerized for 1 h suggests that the limited availability of DA during the short polymerization period resulted in inadequate PDA coating. This coating deficiency may be insufficient to protect MLs from the aggregation caused by DA acting as a salt after dialysis and multiple centrifugation rinses. At high magnifications, the PDA-ML sample observed after 1 h of polymerization (Figure B) reveals gaps (indicated by red circles), suggesting that an incomplete PDA layer had formed during the short polymerization period. PDA-ML samples observed at different polymerization times showed distinct characteristics. The PDA-MLs after 4 h of polymerization exhibited a thicker layer of approximately 6 nm (Figure D), with some vesicles displaying an irregular PDA coating on their surfaces (Figure C). In contrast, PDA-ML samples after 24 h of polymerization possessed a significantly thicker PDA layer of approximately 10 nm (Figure E), and many were fully encapsulated in PDA (Figure F). Additionally, minor impurities observed in the TEM images were identified as nonmagnetic, self-polymerized dopamine residues that do not contribute to the measured signal. TEM images showed that the sizes of PDA-MLs at 1, 4, and 24 h were 68.1 ± 0.2, 76.2 ± 5.4, and 110.4 ± 0.1 nm, respectively. These sizes are considerably smaller than the hydrodynamic size measured by DLS, suggesting that only a small fraction of the sample formed larger MLs. Based on the amounts of MNPs, lipids, and the resulting liposomes, we estimated that each liposome encapsulated approximately five MNPs (detailed calculation in Appendix 1). The magnetization behavior of MLs changes after PDA modification, as the magnetic response is influenced by the Brownian relaxation of the entire PDA-ML structure. Thicker PDA coatings increase the hydrodynamic size of MLs, thereby shifting the characteristic frequency of Brownian relaxation to lower frequencies. One limitation of this work is the lack of stability testing for MLs in serum or plasma, which would be necessary to assess their performance in complex biological environments.

1. Hydrodynamic Size and Zeta Potential of PDA-MLs in Water .

Polymerization time (h)	Z-average (nm)	Zeta potential (mV)
1	178.9 ± 3.0	–37.4 ± 2.2
4	190.6 ± 3.4	–33.5 ± 6.4
24	204.5 ± 5.0	–23.1 ± 1.0

^aThe reported data represent the mean ± standard deviation (SD) from three independent measurements.

3.

Representative low-magnification (A, C, and E) and high-magnification (B, D, and F) TEM images of the collection of PDA-ML prepared using three different polymerization times. (A–B) 1 h, (C–D) 4 h, and (E–F) 24 h.

Lectin–Glycan Detection Based on Brownian-Relaxation Behavior

In this study, we selected specific glycan types that could serve as appropriate controls with well-characterized carbohydrate–lectin interactions. We examined the lectin–glycan interaction using two monoglycans, one terminated with Gal-N-glycan and the other with Sia-N-glycan, which specifically bind to PNA and SNA lectins, respectively. PNA belongs to the galectin family of lectins, which are β-galactoside-binding proteins that specifically recognize and bind to terminal Gal-β­(1,3)-GalNAc residues. Conversely, SNA lectin exhibits high specificity for glycans that terminate with sialic acid (Sia), typically Neu5Ac, in an α­(2,6) linkage.

Figure summarizes the frequency spectra of the AC magnetization for mixtures of lectin-immobilized PDA-MLs and monoglycans. The polymerization time for the PDA coating was 1 h. Figure A and B show the frequency spectra of the real and imaginary parts of the AC magnetization using PNA-immobilized PDA-MLs, whereas Figures C and D show the spectra using SNA-immobilized PDA-MLs. The frequency spectra of the real part in the absence of glycans and for a mismatched lectin–glycan pair are similar. In contrast, the presence of a specifically interacting glycan resulted in a significant decrease in the real part of the AC magnetization; a small peak shift was observed in the imaginary part (Figures B and D). The maximum value in each imaginary spectrum corresponds to the peak of the normalized frequency spectra. The peak frequency was determined as the frequency at which the first peak emerged when scanning from the high-frequency end toward lower frequencies because the spectrum was noisy below 30 Hz. The peak frequency in the spectrum of the imaginary part theoretically decreases with increasing hydrodynamic size of the magnetic particles in solution, which reflects clustering and aggregation.

4.

Frequency spectra of AC magnetization for lectin-immobilized PDA-MLs and monoglycan mixtures. PNA-ML = PNA lectin-immobilized PDA-ML; SNA-ML = SNA lectin-immobilized PDA-ML. (A, B) Real and imaginary parts for PNA-ML, PNA-ML–Sia-N-glycan, and PNA-ML–Gal-N-glycan. (C, D) Real and imaginary parts for SNA-ML, SNA-ML–Gal-N-glycan, and SNA-ML–Sia-N-glycan. The polymerization time for the PDA coating was 1 h.

Because lectin–monoglycan interactions are likely one-to-one, monoglycans theoretically do not facilitate strong cross-linking between lectin-immobilized MLs. However, we observed a significant reduction in the real part only for specific lectin–glycan pairs (Figure A,C). This suggests that glycan binding alters ML surface properties, enabling some MLs to form clusters, which promotes aggregation and subsequent precipitation at the bottom of the sample tubes. This precipitation, in turn, reduces the AC magnetization of the remaining supernatant. In the imaginary part of the data, a small but noticeable reduction in peak frequency was observed for specific lectin–glycan pairs (Figures B and D and Table S1), reflecting the hydrodynamic behavior of nonprecipitated MLs dispersed in the supernatant. This clustering or aggregation is likely caused by the interaction between functional groups such as the COOH groups, which are present at one end of the monoglycan chain and become exposed on the surface when monoglycans bind to MLs, and the lipid molecules of another ML, thereby inducing inter-ML interactions and altering the hydrodynamic behavior. Since it is presumed that the PDA coating is insufficient, as mentioned above, allowing lipid molecules to remain exposed on the ML surface, electrostatic interactions or hydrogen bonding, such as COO^– with N⁺ in DPPC/DSPC/DSPE-PEG, and −OH groups (from glycan hydroxyls) forming H-bonds with phosphate groups in lipid molecules, could be induced. Because individual lectin–glycan interactions are generally weak and reversible, partial precipitation and subtle hydrodynamic modulation of MLs, detected magnetically as shown in Figure , occurred without severe precipitation.

Figure shows the hydrodynamic size of the lectin-immobilized PDA-MLs and monoglycan mixtures measured using DLS. The reduction in peak frequency for PNA-immobilized PDA-MLs (Figure B) agrees with the hydrodynamic-size increase presented in Figure A. However, we observed differences in the hydrodynamic-size trends between PNA- and SNA-immobilized PDA-MLs in the case of the specific lectin–glycan interaction, as shown by the DLS data in Figure . Unlike PNA-immobilized PDA-MLs (Figure A), the hydrodynamic sizes remained similar in all cases for SNA-immobilized PDA-MLs (Figure B). These findings suggest that the interaction between PNA and Gal-N-glycan is stronger than that between SNA and Sia-N-glycan. This difference is supported by the inherent properties of PNA, which forms stable hydrogen bonds and van der Waals forces with the β-D-galactose residue at its binding site. In contrast, Sia is negatively charged, which contributes to a weaker electrostatic interaction. Additionally, the structure of galactose in galactosylated N-glycan is simpler than that of Sia, reducing steric hindrance and enabling optimal interaction. The stronger interaction between PNA lectin and Gal-N-glycan is therefore expected to result in a greater number of glycans binding to PNA on MLs compared with the SNA–Sia-N-glycan case. These additional glycans, in turn, exhibit higher affinity to other MLs via the mechanism described above, facilitating clustering. This explains the hydrodynamic size difference between PNA- and SNA-immobilized MLs interacting with specific glycans, as shown in Figure .

5.

Hydrodynamic sizes of the lectin-immobilized PDA-MLs and monoglycan mixtures were measured using DLS. PNA-ML = PNA lectin-immobilized PDA-ML; SNA-ML = SNA lectin-immobilized PDA-ML. (A) PNA-ML, PNA-ML–Sia-N-glycan, and PNA-ML–Gal-N-glycan. (B) SNA-ML, SNA-ML–Gal-N-glycan, and SNA-ML–Sia-N-glycan. The polymerization time for the PDA coating was 1 h.

A small but noticeable reduction in peak frequency was observed for both cases in the AC magnetization measurements (Figure ). We believe that AC magnetization measurements in solutions detect not only changes in hydrodynamic diameter owing to aggregation but also weak and reversible inter-ML interactions mediated by interactions with surrounding glycans, as well as changes in effective viscosity. Our previous work demonstrated that AC magnetization spectra were modulated through interparticle or intercluster interactions in MNP suspensions. These results indicate that AC magnetization measurements offer higher sensitivity for detecting weak and reversible glycan–lectin interactions.

The size of a monoglycan is only a few nanometers (less than 5 nm); thus, we did not observe a significant peak shift in the frequency spectrum of the AC magnetization. Therefore, glycans do not effectively act as cross-linkers to facilitate aggregate formation. In addition, steric factors and lectin orientation inevitably contribute to binding, as poor orientation or steric crowding lowers the probability of effective interactions, leading to weak or no aggregation. The degree of aggregation depends on how many lectin-carbohydrate-recognition domains (CRDs) can simultaneously engage with glycans on the same or different MLs. Both PNA and SNA are multivalent lectins. For aggregation or cluster formation to occur, at least two binding sites on a lectin must align properly with the glycans. In our case, either PNA or SNA lectin was immobilized on the PDA-ML surfaces. When lectins are adsorbed or conjugated in an orientation in which their binding pockets face away from glycans (e.g., blocked or shielded by the ML surface), binding efficiency is impaired. We expected PDA-mediated binding to promote a more favorable orientation, ensuring CRD exposure and enhancing binding activity, while random adsorption often causes the partial inactivation of lectins.

For further analysis, we employed multivalent glycan-modified human serum albumin (HSA; N-glycoalbumin) and examined the behavior of MLs under lectin concentration-dependent conditions. The lectin concentration refers to the concentration of lectins during the immobilization process, which influences the density of lectins on MLs. Strong binding interactions are favorable in the presence of glycocluster structures. With matching glycan patterns and lectin expressions, glycoclusters can exhibit robust and selective cell interactions. We specifically selected galactosylated N-glycoalbumin (Gal-N-glycoalbumin) and α­(2,6)-sialylated N-glycoalbumin (Sia-N-glycoalbumin); these two types of N-glycoalbumin specifically bind to PNA and SNA, respectively.

Our investigation into interactions between lectins (receptors) and multivalent glycans (analytes), using various polymerization times and lectin concentrations, revealed the complex nature of aggregate formation in lectin-immobilized PDA-MLs with N-glycoalbumin. Shorter polymerization times result in thinner PDA layers, which may leave some areas of the ML surface exposed to its surroundings owing to incomplete coating. This can reduce the effectiveness of lectin immobilization owing to the lower availability of functional groups necessary for glycan binding or stability. Conversely, prolonged polymerization produces a thicker and more robust PDA layer. This increased thickness provides more functional groups for lectin immobilization on MLs. Consequently, a more substantial PDA coating leads to more severe aggregation, which increases the hydrodynamic size of the lectin-immobilized PDA-ML clusters with glycans. Therefore, we evaluated the formation of ML–glycoalbumin clusters by varying the lectin concentration, which is a part of tuning the performance of engineered MLs. Figure shows the effect of polymerization times and SNA concentrations on the AC magnetization of SNA-immobilized PDA-MLs with Sia-N-glycoalbumin. SNA concentration refers to the amount of SNA used during immobilization to assess its interaction with N-glycoalbumin. Figures A, C, and E illustrate that as the SNA concentration increased, the AC magnetization of MLs decreased, leading to the formation of large clusters, indicated by precipitation. When precipitated, they no longer contribute (or partially contribute) to the AC magnetization. In Figures A and C, with 1 and 4 h polymerization times, a gradual decrease in the AC magnetization (real part) was observed as the SNA concentration increased. The variations in baseline AC magnetization and peak frequency at the 0.00 μg/mL lectin concentration reflect the influence of PDA coating thickness on the hydrodynamic properties of MLs. These baseline differences are inherent to the PDA coating and demonstrate the importance of using appropriate controls for each polymerization condition when evaluating lectin-dependent responses. Cluster formation was further evidenced by significant peak shifts observed at the highest SNA concentration (Figures B and D). This SNA-concentration dependence of the peak shift suggests that shorter polymerization times require more SNA during immobilization to achieve a significant binding efficiency. For comparison, in PDA-MLs with a 24 h polymerization time, a significant decrease in the real part of the AC magnetization was noted at low lectin concentrations together with a significant peak shift in the imaginary part (Figures E and F). Absolute differences of peak shift, −Δf, compared to the reference (at 0.00 μg/mL lectin) were 24.9 ± 8.4 and 25.8 ± 15.8 Hz for 6.25 and 50.0 μg/mL, respectively (Figure F and Table S2). This notable change suggests that PDA-MLs with a 24 h polymerization time possess more robust structures and exhibit more binding sites for lectins, as most MLs are covered with excessively thick PDA layers. This thick coating improved lectin immobilization on PDA-ML surfaces, potentially increasing lectin–N-glycoalbumin binding efficiency. Hence, low lectin concentration is sufficient to provide high lectin density on MLs and achieve maximum binding efficiency with Sia-N-glycoalbumin. No changes in the AC magnetization were observed, even as lectin concentrations increased from the low-concentration range shown in Figures E and F, suggesting that the PDA-ML surfaces were already saturated with lectins at low concentrations.

6.

Frequency spectra of AC magnetization for the SNA-immobilized PDA-ML and glycan mixtures. (A-E) The real part and (E) depict the real part, and (B), (D), and (F) depict the imaginary part. The SNA concentration was 0.00–50.0 μg/mL at different PDA polymerization times of (A–B) 1 h, (C–D) 4 h, and (E–F) 24 h. SNA-immobilized PDA-MLs were mixed with 1 μM of α­(2,6)-sialylated N-glycoalbumin.

Further comparative studies revealed that the cluster formation of PNA-immobilized PDA-MLs with Gal-N-glycoalbumin exhibited a trend of decreasing the real part of the AC magnetization similar to that observed with SNA lectins (Figures A, C, and E). However, no clear peak shift was observed in the imaginary part of the PDA-ML samples with polymerization times of 1 and 4 h, even when the lectin concentration was increased in the low-concentration region (Figures B and D). The reduction in the AC magnetization (real part) indicates that PNA-immobilized PDA-MLs formed aggregates, as evidenced by the noticeable aggregation. However, the clusters that remained in the supernatant appeared to be similar in size to those of untreated MLs (corresponding to 0.00 μg/mL lectin in Figure ). Additionally, the magnetization behavior of PNA-immobilized PDA-MLs with a 24 h polymerization time was comparable to that of SNA-immobilized PDA-MLs, exhibiting a significant decrease and peak shift even at low lectin concentrations, for example, −Δf for 6.25 μg/mL of 30.2 ± 13.0 Hz (Figure F and Table S3).

7.

Frequency spectra of AC magnetization for PNA-immobilized PDA-ML and glycan mixtures. (A-E) The real part and (E) depict the real part, and (B), (D), and (F) depict the imaginary part. The PNA concentration was 0.00–50.0 μg/mL at different PDA polymerization times of (A–B) 1 h, (C–D) 4 h, and (E–F) 24 h. PNA-immobilized PDA-ML were mixed with 1 μM of galactosylated N-glycoalbumin.

A series of experiments involving SNA with Sia-N-glycoalbumin and PNA with Gal-N-glycoalbumin demonstrated that the polymerization time of PDA considerably affects lectin functionalization on MLs. As the polymerization time increases, lectin immobilization on the PDA-MLs also increases, which enhances the interaction efficiency of lectins and facilitates binding with glycans. We chose 1-h polymerization for the monovalent glycan experiment (Figures and ) to establish baseline conditions with an intermediate PDA coating thickness. Our systematic study of polymerization time effects with multivalent glycoalbumins (Figures and ) demonstrated that longer polymerization times enhance lectin immobilization efficiency. For monovalent studies, we expected similar trends with longer polymerization providing more lectin binding sites and potentially stronger responses. However, given the weak nature of monovalent interactions and the small size of the glycans relative to MLs, the improvements are likely to be subtle compared with those observed with multivalent glycoalbumins. One limitation of this work is the lack of quantitative data on lectin loading per PDA-ML. Sensitivity and detection limits should be determined in future work since this study did not investigate the dependence of peak shift on analyte (N-glycoalbumin) concentration. Additionally, calculation of the detection limit was not possible because the standard deviations (SD) for control samples (0.00 μg/mL lectin concentration for both SNA and PNA) were zero due to low spectral resolution in the frequency range below 30 Hz.

Noticeable peak shifts were observed at the highest SNA concentration for SNA-immobilized PDA-MLs interacting with Sia-N-glycoalbumin (Figures A and C); however, in the case of the shorter polymerization times of 1 and 4 h, shown in Figures and , only small peak shifts were observed, even at the highest PNA concentration for PNA-immobilized PDA-MLs interacting with Gal-N-glycoalbumin (Figures A and C). These results indicate that the interaction between SNA and Sia-N-glycoalbumin is stronger than that between PNA and Gal-N-glycoalbumin; this is in contrast to the results observed in experiments using monoglycans (Figures and ). The altered affinity from mono-N-glycan to clustered N-glycan, whereby sialylated glycan exhibited a higher affinity than galactosylated glycans when clustered on albumin, is presumably attributed to the different multivalency effects caused by each N-glycan. The larger number of glycans enhances efficient binding, owing to spatially matched multivalent glycans, thereby improving affinity and specificity. Generally, when spatially matched multivalent binding partners can bind simultaneously, they form a single entity. This leads to the highest enhancement of affinity and selectivity, owing to the most favorable enthalpy and entropy contributions. We estimated the K _D of the interaction between SNA and Sia-N-glycan by accounting for multivalency effects, based on reported K _D values of monovalent interactions. Linman et al. reported an equilibrium K _D of 777 ± 93 nM for SNA binding to Neu5Acα2,6-LHEB, a sialic-acid-terminated lactose derivative, measured by SPR. Regarding multivalency, Dam et al. studied the binding of multivalent glycans to Man/Glc-specific lectin Concanavalin A (ConA) using ITC and found that bi-, tri-, and tetra-valent analogues exhibited 6-, 11-, and 35-fold higher association constant K _a values, respectively. The Sia-N-glycoalbumins used in this study contained nine glycans per albumin. We assumed that approximately 2–3 glycans are involved in the binding between SNA-ML and Sia-N-glycoalbumin, as schematically shown in Figure B. Assuming bivalent binding (6-fold enhancement), the K _D for the interaction between SNA-MLs and Sia-N-glycoalbumins was estimated to be approximately 130 nM (777 nM divided by 6). ,

We conducted control experiments using three polymerization times for ML samples to demonstrate that the lectin–glycan interactions in our study are based on specific relationships. Figure S5 shows the AC magnetization (real and imaginary parts) of PDA-MLs with polymerization times of (A–B) 1 h, (C–D) 4, and (E–F) 24 h. PDA-MLs without lectins did not exhibit clear changes in the AC magnetization after mixing with glycans. Importantly, a slight decrease in the AC magnetization indicated a weak interaction arising from the mismatched binding of PNA with Sia-N-glycoalbumin and SNA with Gal-N-glycoalbumin; however, the clear changes seen in Figures , , and did not occur. This confirms that mismatched binding inhibited the aggregation of PDA-MLs.

Conclusions

We developed a multifunctional nanocomposite system comprising PDA-MLs that demonstrates an alternative glycoengineering approach for investigating interactions between lectins and multivalent N-glycoalbumin ligands using magnetic signals derived from the Brownian-relaxation behavior of MLs. By systematically controlling the PDA polymerization time (1–24 h), we achieved the precise tuning of ML surface properties, with extended polymerization producing robust, biocompatible coatings that enhance lectin-immobilization efficiency and structural stability. The engineered PDA-ML nanocomposites exhibited remarkable multivalency effects in lectin–glycan recognition, demonstrating a substantially enhanced binding affinity with N-glycoalbumin compared with that of monovalent saccharide Gal-N-glycan and Sia-N-glycan. N-glycoalbumin can act as a cross-linker in lectin–glycan binding, facilitating the cluster or aggregate formation of MLs. Severe aggregation occurred only when specific lectin–glycan combinations were present, which was indicated by a decrease in the real part of AC magnetization. As more aggregates formed, the AC magnetization decreased due to significant precipitation. A significant decrease in peak frequency was observed as the hydrodynamic size of the MLs increased. By controlling the presentation of sugars on ML surfaces, we established a simplified model of cellular interactions. These modifications bestow biocompatibility on MLs and enhance their capacity for the facile adsorption of biomolecules as well as effective intracellular delivery, paving the way for potential biomedical applications. These findings have significant implications for the development of innovative biomedical solutions and the understanding of lectin–glycan interactions. Future studies should focus on increasing the ML concentration to mitigate the low-frequency noise in AC magnetization, especially after lectin–glycan binding, leading to severe aggregation and subsequently lowering of the AC magnetization signal. Moreover, investigating the effects of different ions and salt concentrations on the binding activity of lectin–glycan and the stability of MLs could offer further insights into optimizing interaction efficiency.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.5c12462.

• Schematic illustration of double-emulsion method for synthesis of magnetoliposomes (MLs); Characterization of iron oxide nanoparticles (IONPs) modified with either APS or CA and MLs using TEM, DLS, zetasizer, XRD, and ATR-FTIR; AC magnetization data for liquid and solid MNPs; Peak frequencies for a series of MLs and glycans; Peak frequencies obtained from SNA-ML–Sia-N-glycoalbumin and PNA-ML–Gal-N-glycoalbumin (polymerization time: 24 h); Control experiment for nonspecific interactions between polydopamine (PDA)-modified MLs with PNA and SNA lectins; Appendix 1. Calculation of the number of encapsulated MNPs (PDF)

¶.

A.N.R. and K.O. contributed equally to this paper. All authors contributed to the writing of this manuscript and have approved the final version for submission. A.N.R.: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Funding acquisition, Writing – original draft, Writing – review and editing. K.O.: Formal analysis, Investigation, Visualization, Funding acquisition, Writing – original draft, Writing – review and editing, Supervision. A.N.: Investigation. T.-C.C.: Funding acquisition, Writing – review and editing. K.T.: Funding acquisition, Writing – review and editing. Y.K.: Funding acquisition, Project administration, Writing – review and editing, Supervision.

The authors declare no competing financial interest.