Fabrication of vitamin B₁₂-loaded multivesicular liposomes via microfluidic double-emulsification: Extended pharmacokinetics and bioavailability

Abstract

Vitamin B₁₂ (VB₁₂) is essential for treating pediatric methylmalonic acidemia (MMA) by activating methylmalonyl-CoA mutase and reducing toxic metabolite levels. However, its clinical utility is constrained by frequent intramuscular injections (leading to poor compliance) and the low, highly variable bioavailability associated with oral administration. Thus, a long-acting, sustained-release VB₁₂ system is urgently needed. Herein, VB₁₂-loaded multivesicular liposomes (VB₁₂-MVLs) were reported for the first time, prepared via an improved microfluidic-double emulsification method. To overcome the loading limitation imposed by VB₁₂'s water solubility, dimethyl-β-cyclodextrin (DM-β-CD) was introduced to construct VB₁₂-β-MVLs. The resulting VB₁₂-MVLs exhibited a characteristic multi-chambered “honeycomb” structure with uniform size (D₅₀ ∼ 14.30 μm) and high encapsulation efficiency (92.43%). DM-β-CD significantly increased the drug concentration from 10.96 to 17.60 mg mL⁻¹. In vivo studies showed that the lipid phase remained at the subcutaneous injection site for >4 days, while the aqueous phase sustained release for up to 7 days. Pharmacokinetic analysis revealed that compared with free VB₁₂ (mean residence time, MRT = 1.50 h), VB₁₂-MVLs extended MRT to 80.69 h, and VB₁₂-β-MVLs further extended it to 121.42 h, with a relative bioavailability of 191.89%. Preliminary safety evaluations indicated no hemolysis, minimal tissue irritation, and no significant changes in hematological or serum biochemical indices. Collectively, the VB₁₂-MVLs platform—particularly VB₁₂-β-MVLs formulation—represents a promising long-acting formulation for MMA, with the potential to reduce dosing frequency and improve the quality of life of pediatric patients while maintaining therapeutic efficacy.

Graphical Abstract

Highlights

• •MVL-based long-acting delivery system for vitamin B₁₂ with cyclodextrin integration.
• •Microfluidic method produces uniform MVLs with 92.43% encapsulation efficiency.
• •Non-classical “up-and-down” binding mode of DM-β-CD with VB₁₂ revealed by docking.
• •VB₁₂-β-MVLs extend MRT to 121.4 h with 191.89% relative bioavailability.
• •Safe, child-friendly formulation enabling 10–14 day dosing interval for MMA patients.

1. Introduction

Methylmalonic acidemia (MMA) is a severe autosomal recessive metabolic disorder associated with impaired methylmalonyl-CoA mutase activity or vitamin B₁₂ (VB₁₂)-related metabolism, and is characterized by substantial morbidity, mortality, and poor long-term prognosis (Almási et al., 2019; Forny et al., 2021). Screening data from more than 7.8 million newborns in China indicate an incidence of approximately 1/15,213, with the VB₁₂-responsive cblC type being the most common (Du et al., 2023). Regular parenteral VB₁₂ supplementation is currently a standard treatment for responsive MMA because it can enhance residual enzyme activity and reduce the accumulation of toxic metabolites (Yu et al., 2021). However, long-term intramuscular administration imposes a substantial treatment burden, particularly in pediatric patients, and may compromise adherence (Castelli et al., 2011). In addition, free VB₁₂ is rapidly cleared after injection, resulting in pronounced fluctuations in systemic drug concentrations (Butler et al., 2006; Martínez-Navarrete et al., 2026; Zhou, 2023). Oral administration is also limited by the saturable intrinsic factor- and ileal receptor-mediated absorption pathway, leading to low and variable bioavailability (Mao et al., 2026; Watanabe, 2007; Wei et al., 2025). Therefore, developing a novel VB₁₂ delivery system that can achieve long-acting sustained release and reduce injection frequency is of considerable clinical value.

Multivesicular liposomes (MVLs), represented clinically by the DepoFoam® platform, are lipid-based depot systems composed of multiple non-concentric aqueous compartments separated by lipid bilayers (Angst and Drover, 2006; Li et al., 2024; Sreelaya and Bhattacharya, 2023; Chaurasiya et al., 2022). This multivesicular architecture provides greater structural stability than conventional unilamellar or multilamellar liposomes and allows drug release to proceed gradually rather than through an “all-or-nothing” process following vesicle disruption (Li et al., 2019; Manna et al., 2019). The relatively large internal aqueous volume also enables efficient encapsulation of hydrophilic compounds, while the lipid components provide a biocompatible basis for long-acting delivery. Several MVL-based products, including Exparel® and DepoCyt®, have been used clinically, supporting the feasibility of this delivery platform (Angst and Drover, 2006; Bulbake et al., 2017; Li et al., 2024). These characteristics make MVLs an attractive system for sustained VB₁₂ delivery. However, although hydrophilic drugs can achieve high encapsulation efficiency (EE) in MVLs, the absolute concentration of VB₁₂ that can be incorporated into the internal aqueous compartments remains constrained by its limited aqueous solubility.

Cyclodextrins can improve the apparent solubility of drugs through non-covalent complexation involving their hydrophobic cavities (Abdelkader et al., 2025; Jiang et al., 2011; Khatoon et al., 2025; Li et al., 2026; Nicolaescu et al., 2025; Yoshikawa et al., 2026). This provides a potential strategy for addressing the concentration limitation of VB₁₂ within the aqueous compartments of MVLs. We therefore hypothesized that pre-complexing VB₁₂ with dimethyl-β-cyclodextrin (DM-β-CD) before incorporation into MVLs could increase the achievable drug concentration without compromising the characteristic multivesicular structure, while potentially providing an additional means of modulating drug release. This formulation strategy addresses a specific question that remains relevant to long-acting VB₁₂ delivery: whether cyclodextrin-assisted solubilization can alleviate the drug-concentration limitation of VB₁₂ while preserving the structural and sustained-release properties of the MVL system.

In this study, VB₁₂-loaded multivesicular liposomes (VB₁₂-MVLs) and a DM-β-CD-assisted formulation (VB₁₂-β-MVLs) were developed for the first time using an improved microfluidic-double emulsification method (Fig. 1). Formulation composition and preparation parameters were optimized, followed by systematic evaluation of physicochemical characteristics, in vitro release and stability, in vivo distribution, pharmacokinetics, and preliminary safety. This study aims to provide a long-acting treatment option for MMA that can substantially reduce dosing frequency.

Fig. 1.

Schematic diagram of the preparation process of VB₁₂-MVLs and VB₁₂-β-MVLs using an improved microfluidic-double emulsification method.

2. Materials and Methods

2.1. Materials

Unless otherwise stated, all reagents and materials were of analytical grade. Phospholipids LipA (phosphatidylcholine, Lot#: 556610-2220019-01) and LipB (charged phospholipid, Lot#: 560300-2230054-01) were purchased from Lipoid GmbH (Germany); LipC (cholesterol, Lot#: B90867) from Wako Pure Chemical Industries (Japan); and LipD (triolein, Lot#: MKCQ) from Sigma-Aldrich (USA). VB₁₂ was obtained from Meryer (China), l-lysine from TCL, and glucose from Thermo Fisher. Dimethyl-β-cyclodextrin (DM-β-CD, molecular weight 1331.36 g/mol, average degree of substitution 11.2–13.3), hydroxypropyl-β-cyclodextrin (HP-β-CD, molecular weight ∼ 1541 g/mol, average degree of substitution 3–6), and sulfobutyl-β-cyclodextrin (SBE-β-CD, molecular weight ∼ 1451–2163 g/mol, average degree of substitution 6.2–6.9) were purchased from Damas Beta (China). The fluorescent dye DiD (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine) and FITC (fluorescein isothiocyanate) was obtained from Bioqueast and Beijing Lanjieke Technology Co., Ltd., respectively, and indocyanine green (ICG) from Aladdin Biochemical Technology Co., Ltd. Methanol was of chromatographic grade (TEDIA). Double-distilled water was used throughout.

2.2. Experimental animals

Female BALB/c mice (16–20 g) and healthy male Sprague–Dawley (SD) rats (200–250 g) were obtained from the Laboratory Animal Center of Hebei Medical University (Shijiazhuang, China). Female C57 mice (16–20 g) were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (China). All animals were housed under controlled conditions at 25 ± 2 °C and 50–75% relative humidity with a 12-h light/dark cycle and had free access to food and water. All animal care and experimental procedures were conducted in accordance with the ARRIVE 2.0 guidelines and were approved by the Laboratory Animal Ethical and Welfare Committee of Hebei Medical University (Approval No. IACUC-Hebmu-2024057).

2.3. Preparation of MVLs and investigation of formulation factors

VB₁₂-MVLs were prepared using a modified double-emulsion method combined with microfluidic technology. VB₁₂ was dissolved in double-distilled water to form the internal aqueous phase (W1), whereas LipA, LipB, LipC, and LipD at the specified formulation ratio were dissolved in a chloroform–methanol–ether mixture to form the organic phase (O). Both phases were preheated to 45 °C and introduced into a T-junction microfluidic system constructed from two polytetrafluoroethylene (PTFE) tubes connected by a T-shaped connector. The phases were mixed at a flow rate of 3 mL/min in the dripping regime and subsequently subjected to ultrasonication at 150 W for 1 min using an ultrasonic cell disruptor to generate the W1/O primary emulsion. No additional surfactant was added to the primary emulsion. Immediately after ultrasonication, the emulsion was cooled on ice and transferred into the external aqueous phase (W2), which contained PVP, glucose, and l-lysine. For secondary emulsification, the W1/O primary emulsion was introduced into W2 through the T-junction microfluidic system under magnetic stirring at 1000 rpm to produce the W1/O/W2 double emulsion. The total flow rate (TFR) and flow-rate ratio (FRR) were optimized, and the final preparation conditions were a TFR of 3 mL/min and an FRR of 1:1. The resulting double emulsion was diluted with 0.9% NaCl, followed by removal of the organic solvents by rotary evaporation at 37 °C. The MVL suspension was then centrifuged at 7000 rpm for 10 min and washed three times with 0.9% NaCl to obtain VB₁₂-MVLs (Mu et al., 2018).

For the preparation of VB₁₂-β-MVLs, VB₁₂ and DM-β-CD at the specified molar ratio were mixed in 1 mL of double-distilled water and sonicated at 300 W for 5 min to form the VB₁₂-β-CD complex. The resulting solution was used as the internal aqueous phase, and the remaining preparation steps were identical to those described for VB₁₂-MVLs.

Formulation factors were investigated using single-factor experiments. Encapsulation efficiency (EE), drug loading (DL), particle size, and other formulation-specific parameters were used as evaluation indices. The LipA:LipC ratios evaluated were 1:2, 1:1, 3:2, and 2:1; LipD concentrations were 1, 2, and 3 mg mL⁻¹; and l-lysine concentrations were 20, 40, and 60 mmol/L.

The effects of osmotic-pressure conditions were evaluated by varying the type and concentration of osmotic-pressure modifiers in the internal and external aqueous phases while keeping the other formulation parameters constant. The osmotic pressures of the different solutions were measured using a freezing-point osmometer. The internal aqueous phases included water, 1% sucrose, 3% sucrose, and 1% glucose, whereas the external aqueous phases included 2%, 3.2%, and 5% glucose. Encapsulation efficiency, leakage rate (LR), sedimentation volume ratio (F), redispersibility, and syringeability were evaluated.

The sedimentation volume ratio was determined according to the suspension section of the Chinese Pharmacopoeia. The initial suspension height was recorded as H₀, and the height after settling was recorded as H; F was calculated according to Eq. (1):

$$F=\left(\frac{H}{H_0}\right)\times 100\%$$ (1)

After sedimentation, redispersibility was evaluated by inverting the vial 20 times/min and observing the extent and rate of redispersion. Syringeability was assessed after vigorous shaking by drawing the suspension into a 2.5-mL syringe fitted with 3-, 4.5-, or 5-gauge needles and recording the time required to withdraw 1 mL.

Aqueous to organic phase volume ratio: Keeping other factors constant, the effect of volume ratios (2:1, 1:1, 1:1.5, 1:2, 1:4) on EE, DL, ζ potential, and 7 day LR was evaluated.

2.4. Investigation of preparation process parameters

The effects of primary-emulsion ultrasonication and microfluidic flow rate on the preparation of VB₁₂-MVLs were investigated while the other formulation and process parameters were kept constant.

For optimization of the primary W1/O emulsion, ultrasonication powers of 60, 100, 150, and 200 W were evaluated at a fixed sonication time of 1 min. The effects of ultrasonication power on the median particle diameter (D₅₀), mean particle size, and particle count of the resulting MVLs were assessed.

To investigate the effect of microfluidic mixing conditions, flow rates of 1.5, 3, and 6 mL/min were evaluated while the other preparation parameters were maintained constant. Encapsulation efficiency (EE), drug loading (DL), and particle size distribution were determined for the resulting MVLs.

2.5. Optimization of VB₁₂-to-DM-β-CD molar ratio and VB₁₂-β-MVLs

HP-β-CD, DM-β-CD, and SBE-β-CD were separately used to prepare VB₁₂-β-MVLs according to the procedure described above. Encapsulation efficiency (EE), 7-day leakage rate (LR), and microscopic morphology were compared to screen the cyclodextrin type.

After selection of DM-β-CD, the effect of the VB₁₂:DM-β-CD molar ratio was investigated while the other preparation conditions were kept constant. Molar ratios of 2:1, 1:1, and 1:2 were evaluated based on EE, drug concentration, and particle size of the resulting VB₁₂-β-MVLs.

The ultrasonication conditions used for formation of the VB₁₂-β-CD complex were subsequently optimized. The conditions evaluated were 100 W for 5 min, 150 W for 5 min, 300 W for 5 min, and 150 W for 30 min. The effects of these conditions on EE, drug concentration, and particle size of the resulting VB₁₂-β-MVLs were determined.

2.6. Characterization and physicochemical properties of MVLs

Morphological morphology: The color, dispersibility, and sedimentation of the MVLs suspension were visually assessed. For transmission electron microscopy (TEM), samples were negatively stained with 2% phosphotungstic acid and examined using an H-7800 transmission electron microscope (Hitachi). For confocal laser scanning microscopy (CLSM), dual-labeled MVLs were prepared with FITC in the aqueous phase (excitation/emission, 488/525 nm) and DiD in the lipid phase (excitation/emission, 646/663 nm) and examined using an LSM900 confocal microscope (Zeiss).

Particle size and ζ-potential. Particle size was measured using a Coulter counter, with three measurements performed for each sample. The ζ-potential was determined using a Malvern Zetasizer, also with three measurements per sample.

Determination of VB₁₂ content, EE, DL, and LR. After rotary evaporation, the volume (V) of the MVL suspension was recorded. To determine the total VB₁₂ concentration, 100 μL of unpurified MVLs was diluted as appropriate and disrupted with methanol at a ratio of 1:3. To determine the free VB₁₂ concentration, the remaining MVL suspension was centrifuged at 7000 rpm for 10 min, and 100 μL of the supernatant was collected. VB₁₂ concentrations were quantified by high-performance liquid chromatography (HPLC).

The detection wavelength was determined by scanning VB₁₂ solutions over 200–600 nm. HPLC analysis was performed using a Kromasil C18 column (150 × 4.6 mm, 5 μm) with methanol–0.02% phosphoric acid (30:70, v/v) as the mobile phase at a flow rate of 0.7 mL/min. The column temperature was maintained at 40 °C, the detection wavelength was 360 nm, and the injection volume was 20 μL.

The HPLC method was evaluated for specificity, linearity, recovery/accuracy, precision, repeatability, limit of detection (LOD), and limit of quantification (LOQ). Specificity was assessed by comparing chromatograms of VB₁₂ solution, blank liposomes disrupted with Triton X-100, and blank DM-β-CD-containing liposomes. For linearity assessment, VB₁₂ standards at 2.5, 5, 10, 20, 40, 50, 60, 80, and 100 μg/mL were analyzed in triplicate, and peak area was plotted against concentration. Recovery/accuracy was evaluated at 80, 100, and 120 μg/mL. Intra-day precision was assessed at five time points within one day, whereas inter-day precision was evaluated once daily for three consecutive days at the same concentration levels. Repeatability was assessed by six consecutive injections of the 100 μg/mL solution. For LOD and LOQ, we determined these values by the signal-to-noise (S/N) approach using serial dilutions of the VB₁₂ standard solution. The concentration that gave S/N ≈ 3 was taken as the LOD, and the concentration with S/N ≈ 10 was taken as the LOQ. The measured values are:  LOD = 0.8 μg/mL (S/N ≈ 3); LOQ = 2.5 μg/mL (S/N ≈ 10). Notably, the LOQ coincides with the lowest concentration of the calibration curve (2.5 μg/mL), which was already included in the linearity assessment and showed excellent precision and accuracy. This confirms that the LOQ is fully adequate for quantifying VB₁₂ in the liposomal formulations. EE, LR, and DL were calculated using the following formulas:

$$\mathrm{EE}=\frac{{\mathrm{C}}_{\text{totol}}-{\mathrm{C}}_{\text{free}}}{{\mathrm{C}}_{\text{totol}}}\times 100\%$$ (2)

$$\mathrm{LR}=\frac{{\mathrm{EE}}_0-{\mathrm{EE}}_7}{{\mathrm{EE}}_0}\times 100\%$$ (3)

$$\mathrm{DL}=\frac{{\mathrm{C}}_{\text{encapsulated}}\times \mathrm{V}}{\left({\mathrm{C}}_{\text{encapsulated}}\times \mathrm{V}+\text{mass of lipids in MVLs}\right)}\times 100\%$$ (4)

where EE₀ is the initial EE and EE₇ is the EE after 7 days.

Fourier transform infrared spectroscopy and X-ray diffraction. For Fourier transform infrared (FTIR) spectroscopy, approximately 1 mg of each test sample, including drug, blank MVLs, and drug-loaded MVLs in freeze-dried form, was mixed with approximately 160 mg of KBr, ground, and compressed into pellets. Spectra were recorded over 400–4000 cm⁻¹. For X-ray diffraction (XRD), the drug, blank MVLs, and freeze-dried drug-loaded MVLs were placed on a glass plate and analyzed using Cu Kα radiation at 40 kV and 30 mA. Diffraction patterns were recorded over 5°–40° with a step size of 0.02° and a scan rate of 2°/min.

In vitro release study: Dialysis membranes with a molecular-weight cutoff (MWCO) of 8000–14,000 Da were cut into 5–7 cm segments and boiled in 2% NaHCO₃ containing 1 mmol/L EDTA·2Na (pH 8.0) for 30 min. The membranes were rinsed with distilled water, boiled again in 1 mmol/L EDTA·2Na (pH 8.0) for 30 min, cooled, and stored in distilled water at 4 °C until use.

For determination of the initial VB₁₂ concentration, 100 μL of purified VB₁₂-MVLs was diluted with 5 mL of 0.9% NaCl, treated with 200 μL of Triton, and analyzed by HPLC. A defined volume of MVLs was transferred into a dialysis membrane(MWCO 8000–14,000) and immersed in 10 mL of PBS (pH 7.4). Samples were incubated at 37 °C with shaking at 100 rpm. At predetermined time points (0, 5, 10, 24, 30, 42 h, etc.),1 mL of release medium was withdrawn and replaced with an equal volume of fresh medium. VB₁₂ concentrations in the collected samples were determined by HPLC, and the cumulative release (Q) of VB₁₂ was calculated according to the formula (Mu et al., 2018). The in vitro release curve of VB₁₂ was plotted.

The in vitro release data were fitted to zero-order, first-order, Ritger-Peppas, and Higuchi equations, and the type of drug release was judged based on the Akaike Information Criterion (AIC) and the goodness of fit (R²). The model with the lowest AIC and highest R² was selected as the most appropriate.

The specific equations used for model fitting are as follows:

• (1)Zero-order model: $\mathrm{Q}={\mathrm{k}}_0\cdot \mathrm{t}(\mathrm{or}\mathrm{Y}=\mathrm{a}\cdot \mathrm{t}+\mathrm{b})$, where Q is the cumulative percentage of drug released at time t, and k₀ is the zero-order release rate constant.
• (2)First-order model: ln (1-Q) = −k₁·t or (Q = Q_∞·(1-e^-k₁^t)), where Q_∞ is the maximum cumulative release and k₁ is the first-order rate constant.
• (3)Higuchi model: Q = k_H·t^1/2, where k_H is the Higuchi dissolution constant, describing release as a diffusion process based on Fick's law.
• (4)Ritger-Peppas model: ln Q = ln k_RP + n · lnt, where k_RP is a constant incorporating structural and geometric characteristics of the formulation, and n is the release exponent indicative of the release mechanism (e.g., Fickian diffusion, anomalous transport, or case-II transport).

$$\mathrm{Q}=\frac{{\mathrm{C}}_{\mathrm{n}}{\mathrm{V}}_{\mathrm{n}}+\left({\mathrm{C}}_{\mathrm{n}-1}{\mathrm{V}}_{\mathrm{n}-1}+{\mathrm{C}}_{\mathrm{n}-2}{\mathrm{V}}_{\mathrm{n}-2}+{\mathrm{C}}_{\mathrm{n}-3}{\mathrm{V}}_{\mathrm{n}-3}+\dots +{\mathrm{C}}_1{\mathrm{V}}_1\right)}{\mathrm{M}}$$ (5)

where C_n is the concentration measured by the external standard method at the n-th sampling, V is the total volume of release medium, V_n-1 is the volume withdrawn at the (n-1)-th sampling, and M is the mass of VB₁₂ in the dialysis bag.

Storage and lyophilization stability. VB₁₂-MVLs and VB₁₂-β-MVLs were stored at 4 and 37 °C for 60 days. Samples were collected on days 0, 7, 12, 20, 30, 40, and 60 to evaluate appearance, particle size, EE, and particle count.

For lyophilization stability studies, MVL suspensions were mixed with the following lyoprotectant formulations: 5%, 10%, or 15% mannitol; 1% or 2% sodium alginate; 1% PVA; isotonic 1% PVA; and isotonic 1% sodium alginate. The samples were freeze-dried and subsequently reconstituted, after which morphology and EE were evaluated.

2.7. In vivo distribution study

Fluorescently labeled MVLs were prepared to monitor the in vivo behavior of the internal aqueous and lipid phases. For DiD-labeled MVLs (DiD-MVLs), DiD was incorporated into the organic phase to yield a final concentration of 10 μg/mL in the MVL suspension, and the remaining preparation procedure was identical to that used for VB₁₂-MVLs. For ICG-labeled MVLs (ICG-MVLs), ICG was incorporated into the internal aqueous phase at a concentration of 25 μg/mL, and the remaining preparation steps were identical to those used for VB₁₂-MVLs. The encapsulation efficiency of ICG was determined using a microplate reader, and the particle size of ICG-MVLs was measured.

Female BALB/c mice (16–20 g) were shaved on the dorsal region and randomly assigned to the following groups: a blank control group (n = 2), which received subcutaneous saline; a free ICG group (n = 3), which received 100 μL of ICG solution in saline at 25 μg/mL; and an ICG-MVLs group (n = 3), which received 100 μL of ICG-MVLs containing ICG at 25 μg/mL. An additional DiD-MVLs group (n = 3) was included to monitor the in vivo distribution and persistence of the MVL lipid phase.

At 0, 8, 24, 48, 72, 120, and 168 h after subcutaneous administration, the mice were anesthetized with isoflurane and imaged using a small-animal three-dimensional multimodal imaging system. Fluorescence images were acquired using excitation/emission wavelengths of 640/670 nm for DiD and 785/810 nm for ICG (Xiang et al., 2013). After the final imaging session at 168 h, the mice were euthanized, and the heart, liver, spleen, lungs, and kidneys were collected for ex vivo fluorescence imaging.

2.8. Pharmacokinetic study

Bioanalytical method. A VB₁₂ stock solution (100 μg/mL) was prepared by accurately weighing an appropriate amount of VB₁₂, dissolving it in water, and storing the solution at 4 °C. For plasma sample preparation, 200 μL of plasma was mixed with methanol to precipitate proteins. After centrifugation, the supernatant was collected and concentrated under vacuum to remove methanol. The residue was reconstituted in 200 μL of water and subjected to HPLC analysis.

Chromatographic separation was performed using a Kromasil C18 column (150 × 4.6 mm, 5 μm). The mobile phase consisted of methanol and 0.1% phosphoric acid in water (30:70, v/v) at a flow rate of 0.7 mL/min. The column temperature was maintained at 40 °C, the detection wavelength was 360 nm, and the injection volume was 20 μL. The analytical method was evaluated for specificity, linearity, intra- and inter-day precision, extraction recovery, room-temperature stability, and freeze–thaw stability.

Animal dosing and blood sampling. Male Sprague–Dawley (SD) rats (200–250 g) were randomly assigned to three groups (n = 6 per group): free VB₁₂ solution, VB₁₂-MVLs, and VB₁₂-β-MVLs. All formulations were administered subcutaneously at an equivalent VB₁₂ dose of 20 mg/kg.

Blood samples were collected from the medial canthus before administration (0 h) and at 0.5, 1, 2, 4, 6, 8, 12, 24, 48, 72, 96, and 120 h after dosing. Heparin was used as the anticoagulant. Plasma was separated by centrifugation and analyzed for VB₁₂ using the HPLC method described above.

Pharmacokinetic analysis. Plasma concentration–time data were analyzed by noncompartmental analysis using DAS software (version 3.2.2). The pharmacokinetic parameters included the maximum plasma concentration (C_max), time to maximum plasma concentration (T_max), elimination half-life (t_1/2), mean residence time (MRT), and area under the plasma concentration–time curve from time zero to the last sampling time (AUC_0–t).The relative bioavailability is the ratio of AUC_{0-t (Formulation)} to AUC_{0-t (Free)}.

2.9. Preliminary safety evaluation

In vitro cytotoxicity assay: The cytotoxicity of VB₁₂-MVLs and VB₁₂-β-MVLs was evaluated using mouse fibroblast 3 T3 cells. Cells were seeded in 96-well plates at a density of 5 × 10³ cells per well and cultured overnight at 37 °C in a 5% CO₂ atmosphere. The cells were then treated with various concentrations (1.2, 0.6, 0.3, 0.1, 0.05 and 0.02 mg mL⁻¹) of VB₁₂-MVLs or VB₁₂-β-MVLs for 24 h. Untreated cells served as the negative control (100% viability). After the treatment period, 10 μL of CCK-8 solution was added to each well, followed by incubation for 2 h at 37 °C. The absorbance was measured at 450 nm using a microplate reader. Cell viability was calculated as the percentage of absorbance relative to the negative control. All experiments were conducted in six identical copies.

Histopathological analysis: After the administration period, rats were euthanized, and the heart, liver, spleen, lungs, kidneys, and injection site skin were collected for histopathological examination (Liu et al., 2022; Yang et al., 2026). Tissue samples were fixed in 4% formaldehyde, embedded in paraffin, sectioned, stained with hematoxylin and eosin (H&E), and examined by light microscopy.

Hemocompatibility assay: A 2% rat red blood cell (RBC) suspension was prepared for the in vitro hemocompatibility assessment. MVL samples were tested at final concentrations of 0.5, 1, 2, and 4 mg mL⁻¹. Saline and 1% Triton X-100 were used as the negative and positive controls, respectively. Samples were incubated at 37 °C for 1 and 3 h. Following centrifugation, hemolysis and RBC morphology were assessed visually and microscopically.

Hematological and biochemical evaluation. Blood samples were collected into EDTA-containing tubes, and white blood cell, lymphocyte, neutrophil, and monocyte counts, together with their corresponding percentages, were measured using an automated hematology analyzer. Alanine aminotransferase (ALT), aspartate aminotransferase (AST), urea/blood urea nitrogen (UREA/BUN), and creatinine (CREA) were measured using an automated biochemical analyzer.

2.10. Statistical analysis

Data are expressed as mean ± standard deviation (SD). Student's t-test or one-way analysis of variance (ANOVA) was used, and a P-value<0.05 was considered statistically significant. Statistical significance was denoted as follows: ns (P > 0.05), *P < 0.05, **P < 0.01, and ***P < 0.001.

3. Results

3.1. Formulation and process optimization

Single-factor screening was first performed to optimize the lipid composition and l-lysine concentration of VB₁₂-MVLs. The results indicated that when the LipA: LipC ratio was 1:1, LipD concentration was 2 mg mL⁻¹, and l-lysine concentration was 40 mM, VB₁₂-MVLs achieved the highest EE (87.39 ± 0.85%), DL (23.70 ± 0.54%), and ζ-potential (−51.5 ± 1.8 mV) (Table 1). Microscopic observation showed that the MVLs particles formed under this formulation were spherical (Supplementary Fig. S1). Quantitative particle size analysis by Coulter counter revealed a narrow size distribution with a median diameter (D₅₀) of 14.30 ± 1.98 μm. These results indicate that appropriate lipid ratios and lysine concentration are critical for forming stable MVLs.

Table 1.

The formulation and dosage of each component of MVLs (n = 3, mean ± SD).

	LipA: LipC	LipD (mg mL−1)	Lysine (mM)	EE (%)	DL (%)	Zeta potential (mV)
1	1:2	2	40	85.79 ± 1.24	23.89 ± 0.67	−38.4 ± 2.1
2	1:1	2	40	87.39 ± 0.85	23.70 ± 0.54	−51.5 ± 1.8
3	3:2	2	40	82.23 ± 1.56	17.98 ± 0.72	−49.5 ± 2.3
4	2:1	2	40	80.82 ± 1.87	19.69 ± 0.88	−46.6 ± 2.0
5	1:1	1	40	78.47 ± 2.01	22.41 ± 0.91	−44.7 ± 2.5
6	1:1	3	40	85.99 ± 1.33	21.73 ± 0.76	−41.8 ± 2.2
7	1:1	2	20	72.86 ± 2.14	20.09 ± 0.95	−44.8 ± 2.4
8	1:1	2	60	87.38 ± 0.92	24.51 ± 0.61	−49.4 ± 2.0

The effects of osmotic conditions and phase ratios were then evaluated. The results showed that when the internal water phase in F1 was deionized water (osmotic pressure 0) and the external water phase was 3.2% glucose (osmotic pressure difference 258 mOsmol kg⁻¹), the EE was the highest (93.05 ± 0.16%), and the 7-day LR was relatively low (3.66 ± 0.26%). Under these conditions, the suspension could be withdrawn readily through 0.5- and 0.45-mm needles, whereas withdrawal through a 0.3-mm needle was more difficult (Table S1 and Fig. 2A). Microscopic observations also showed dense, spherical particles under the selected osmotic conditions (Fig. 2B and Supplementary Fig. S2). In the aqueous-to-organic phase ratio screening, a ratio of 1:1 gave the highest EE (85.19%; Table S2). For the drug-to-lipid mass ratio, 1:3 yielded an EE of 93.05 ± 0.16% and a VB₁₂ concentration of 10.96 ± 0.27 mg mL⁻¹, whereas the EE decreased to 79.13% at a ratio of 1:6 (Table S3) (Abuzar et al., 2020).

Fig. 2.

Optimization process of prescription technology for VB₁₂-MVLs and VB₁₂-β-MVLs. (A) The effect of different osmotic pressure regulators on MVLs. (B) Images of samples prepared with different osmotic pressure adjusting solvents taken in upright and inverted positions. (C) Effect of colostrum ultrasonic parameter on EE of MVLs. (D) Effect of mixing flow rate on EE of MVLs. (E-F) Stability of VB₁₂-β-CD solution. (G) The influence of the molar ratio of VB₁₂ to DM-β-CD on the EE of MVLs. (H) The influence of ultrasound power on the EE of MVLs. ns: (P > 0.05), *P < 0.05, **P < 0.01, and ***P < 0.001.

Appropriate ultrasound parameters are crucial for preparing MVLs (Lu et al., 2021). Therefore, further exploration of ultrasound conditions was conducted. The results showed that 150 W 1 min⁻¹ was the optimal primary emulsion ultrasonication parameter (Fig. 2C). At this condition, the EE of MVLs was 92.43 ± 0.314%, DL was 34.05 ± 0.833%, and particle size was moderate (D₅₀ = 14.30 μm). Ultrasonic power directly determined the size of the internal vesicles: the higher the power, the smaller the particle size (Table S4). In addition, when the flow rate of the microfluidic system was 3 mL/min, the EE was highest, at 92.43 ± 0.314%, and the DL was 34.05 ± 0.833% (Fig. 2D). As the flow rate increased, the volume injected into the external aqueous phase per unit time increased while the stirring speed of the external phase was constant, so the MVLs particle size gradually increased (Table S5).

While the VB₁₂-MVLs formulation already achieved a high EE (>90%), the absolute drug concentration (10.96 mg mL⁻¹) remained moderate due to VB12's limited aqueous solubility. To further increase the drug loading capacity—which is critical for reducing injection volume and achieving sustained therapeutic levels—cyclodextrins were introduced. After comparing HP-β-CD, DM-β-CD, and SBE-β-CD, it was found that DM-β-CD had the most significant effect on increasing the solubility of VB₁₂ (Fig. 2E-F). Therefore, DM-β-CD was selected as the solubilizer. When the VB₁₂:DM-β-CD molar ratio was 1:1 and the ultrasonication parameter was 300 W 5 min⁻¹, the drug concentration of VB₁₂-β-MVLs reached 17.60 mg mL⁻¹, the particle size D₅₀ was 16.46 μm, and the EE was 86.19% (Fig. 2G-H, Tables S6 and S7). Deviation from the 1:1 M ratio or excessively long ultrasonication time reduced both EE and DL.

3.2. Physicochemical characterization and formation mechanism

The optimized VB₁₂-MVL and VB₁₂-β-MVL suspensions were dark red. After standing, a light-pink supernatant and dark-red sediment were observed (Fig. 3A and Supplementary Fig. S3A). TEM observation showed that VB₁₂-MVLs were spherical or near-spherical particles with multiple closely packed non-concentric internal chambers (Fig. 3B-D). In addition, confocal laser scanning microscopy (CLSM) (FITC-labeled aqueous phase, DiD-labeled lipid phase) further confirmed the non-concentric multivesicular internal structure of VB₁₂-MVLs and VB₁₂-β-MVLs. The merged images revealed multiple discrete, spatially separated red-fluorescent aqueous compartments, each individually enclosed by a green-fluorescent lipid bilayer, without a common center or concentric arrangement. This pattern is distinct from conventional unilamellar liposomes (single central aqueous core) and multilamellar liposomes (concentric ‘onion-like’ lamellae), thus unambiguously demonstrating the characteristic multivesicular ‘honeycomb’ architecture of the prepared MVLs (Fig. 3E-F and Supplementary Fig. S3B) (Chaurasiya et al., 2022). Notably, VB₁₂-β-MVLs showed no structural difference from VB₁₂-MVLs, indicating that cyclodextrin complexation did not disrupt the typical MVLs architecture. Under the optimal formulation, VB₁₂-MVLs had a favorable particle size, with D₅₀ approximately 14.30 ± 1.980 μm (Supplementary Fig. S4), a uniform size distribution, and a ζ-potential of −51.5 mV. The strong charge repulsion helps electrostatic stabilization of the particles and prevents aggregation (Xiao et al., 2011).

Fig. 3.

Physicochemical characterization and formation mechanism of VB₁₂-MVLs and VB₁₂-β-MVLs. (A) Photographs of VB₁₂-MVLs under uniform dispersion and standing conditions. (B-D) Transmission electron microscopy (TEM) images of VB₁₂-MVLs. The yellow arrows represent multiple, tightly arranged, non concentric inner cavities of the MVL structure. (E-F) Laser confocal scanning microscopy (CLSM) images and 3D results of VB₁₂-MVLs. (G) Fourier transform infrared spectra obtained from different samples. (H) X-ray diffraction analysis patterns obtained from different samples. (I-K) Docking interactions between VB₁₂ molecules and HP-β-CD, SBE-β-CD, and DM-β-CD molecules, respectively. Green represents VB₁₂ molecules. The blue color in (I) represents HP-β-CD; The blue color in (J) represents SBE-β-CD molecule; The blue color in (K) represents DM-β-CD. The yellow dashed line represents hydrogen bond interactions, and the markings indicate hydrogen bond lengths. (L) The molecular docking binding energy of VB₁₂ with various cyclodextrins. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Furthermore, the formation mechanism of MVLs was further studied. FTIR spectra showed that the amide C=O characteristic peak of VB₁₂ (approximately 1670 cm⁻¹) in VB₁₂-MVLs and VB₁₂-β-MVLs, and the hydroxyl peak of DM-β-CD (approximately 3420 cm⁻¹) in VB₁₂-β-MVLs, did not shift or show new peaks, indicating no chemical interaction between the drug and excipients only physical mixing (Fig. 3G). XRD patterns showed that the characteristic sharp peaks of VB₁₂ disappeared in VB₁₂-MVLs and VB₁₂-β-MVLs, and the characteristic sharp peaks of DM-β-CD also disappeared in VB₁₂-β-MVLs, indicating that VB₁₂ and DM-β-CD existed in an amorphous or new solid form within the liposomes, which is beneficial for increasing the dissolution rate (Fig. 3H). Molecular docking is a simulation method that can predict the binding mode between small molecule compounds or large molecules and receptors, and predict their molecular interactions (Pinzi and Rastelli, 2019). Therefore, further molecular docking simulation experiments were conducted. Molecular docking simulation results (Fig. 3I-L) showed that the binding energies of VB₁₂ with HP-β-CD, SBE-β-CD, and DM-β-CD were-2.77, −2.52, and −1.64 kcal mol⁻¹, respectively, all negative, indicating that the binding process can occur spontaneously (Kumar et al., 2024). VB₁₂ adopted an “up-and-down” binding mode with DM-β-CD without entering the cavity, suggesting that the solubilization is primarily mediated by weak interactions such as hydrogen bonding (Jiang et al., 2011). This prediction provides a theoretical basis for the solubilization mechanism of DM-β-CD.

3.3. In vitro release and stability

Next, the in vitro release behavior of MVLs was studied. The results showed that in PBS at pH 7.4, the release behavior of VB₁₂-MVLs was best described by a first-order kinetic model, as it exhibited the lowest AIC value (46.43) and the highest R² (0.9870) among all models evaluated (Table S8) (Shen et al., 2011). Among them, the dosage of LipC (regulating membrane fluidity) and LipD can significantly regulate the release rate. Increasing the ratio of LipC (LipA : LipC = 1:3) can delay release, but excessive (1:2) can accelerate release, which may be related to the increase in membrane rigidity leading to brittle rupture (Fig. 4A) (Holzschuh et al., 2018; Karami and Jalili, 2015). A LipD amount of 4 mg gave the slowest release, while excess (6 mg) might alter the internal aqueous phase pH and accelerate release (Fig. 4B). Notably, the addition of DM-β-CD further delayed the release of VB₁₂ (Fig. 4C), indicating its superior sustained-release potential.

Fig. 4.

In Vitro release and stability of VB₁₂-MVLs and VB₁₂-β-MVLs. (A) In vitro release of VB₁₂-MVLs at different ratios of LipA to LipC (n = 3). (B) In vitro release of VB₁₂-MVLs at different LipD dosages (n = 3). (C) In vitro release of VB₁₂-MVLs and VB₁₂-β-MVLs (n = 3). (D-E) LR of different samples placed at 4 °C and 37 °C. (F-G) Particle size stability of VB₁₂-MVLs and VB₁₂-β-MVLs at 4 °C. (H) Photos of VB₁₂-MVLs after freeze-drying using different freeze-drying protectants. ns: (P > 0.05), *P < 0.05, **P < 0.01, and ***P < 0.001.

Stability is an important indicator for liposomes, so the stability of the samples was further examined. The results showed that after storage at 4 °C for 60 days, the EE of VB₁₂-MVLs and VB₁₂-β-MVLs remained above 72% (Fig. 4D-E), and changes in particle size and particle count were within the specified ranges (Fig. 4F-G), indicating good stability over 60 days. At 37 °C, the EE remained above 80% after 10 days, indicating good short-term thermal stability of the formulation. Lyophilization is a promising method to extend the shelf life of liposomes, but the freezing and drying may cause structural and functional damage to liposomes, leading to leakage of encapsulated drugs (Chen et al., 2010). Therefore, optimizing the lyoprotectant formulation is crucial for stabilizing liposomes. The optimal formulation was diluted with different lyoprotectants, freeze-dried, and the specific morphology after lyophilization was observed (Fig. 4H). The results showed that isotonic 1% sodium alginate provided the best protection, with an EE of 66.71 ± 0.09% after reconstitution (Table S9), and many intact multivesicular structures were still visible under microscopy (Supplementary Fig. S5). Mannitol provided poorer protection, and a high concentration (15%) reduced the EE to 10.97%. Therefore, sodium alginate is a suitable lyoprotectant for MVLs freeze-dried formulations.

3.4. In vivo distribution

The unique micron-scale particle size of MVLs prevents them from entering capillaries, thereby causing them to remain at the administration site. Due to the fact that most of the volume of MVLs is occupied by water, the residual amount of lipids after the release of the contents is very small. Therefore, ICG was used to label the internal aqueous phase of MVLs to study the in vivo release behavior, and DiD was used to label the lipid phase of MVLs to study the subcutaneous residence time of the lipid material (Mu et al., 2018). In vivo imaging results showed that fluorescence in the ICG-MVLs group had essentially disappeared by day 7 (Supplementary Fig. S6). Ex vivo tissue imaging showed that on day 7, fluorescence was still present in the liver of the ICG-MVLs group (Fig. 5A), suggesting that the drug was slowly absorbed and entered the liver for metabolism, confirming the long-acting subcutaneous depot effect of MVLs. After subcutaneous injection of DiD-MVLs, the red fluorescence signal from the lipid phase was mainly distributed at the injection site on the back of the mice and remained at the injection site for more than 4 days (Fig. 5B and Supplementary Fig. S7), indicating that the micron-sized MVLs particles cannot enter capillaries and form a subcutaneous depot, providing a spatial basis for long-acting release (Zhang et al., 2022).

Fig. 5.

In vivo distribution and pharmacokinetics of VB₁₂-MVLs and VB₁₂-β-MVLs. (A) Distribution of ICG in mouse tissues. (B) In vivo imaging results of subcutaneous injection of DiD-MVL in mice. (C) Schematic diagram of plasma pretreatment. (D) Plasma concentration-time profiles of different VB₁₂ formulations within 250 h. The inset shows magnified plasma concentration-time profiles of different VB₁₂ formulations within 25 h. (E) Main pharmacokinetic parameters of VB₁₂ in mice (n = 6). ns: (P > 0.05), *P < 0.05, **P < 0.01, and ***P < 0.001.

3.5. Pharmacokinetic study

Subsequently, SD rats were selected for in vivo pharmacokinetic studies of MVLs, and after administration, plasma samples were pretreated for drug content detection (Fig. 5C). From the concentration-time curves and kinetic parameters, free VB₁₂ administered subcutaneously was rapidly absorbed (C_max = 39.26 μg/mL, T_max = 0.5 h) and quickly eliminated (MRT = 1.50 h). In contrast, VB₁₂-MVLs exhibited a clear sustained-release profile with a double-peak phenomenon (C_max = 1.44 μg/mL, T_max = 48.0 h), and the MRT was prolonged to 80.69 h, with a relative bioavailability of 97.02% (Fig. 5D-E). Compared with VB₁₂-MVLs, VB₁₂-β-MVLs had a T_max of 60.5 h, an MRT further prolonged to 121.42 h, an AUC_0-t increased to 119.72 h·μg/mL, and a relative bioavailability of 191.89%, approximately twice that of VB₁₂-MVLs. This indicates that DM-β-CD not only increases drug loading but also optimizes the in vivo behavior of the drug, achieving long-acting sustained release for more than 10 days in rats. Therefore, MVLs are an effective controlled-release carrier with clinical potential (Luo et al., 2016), which is especially important for children with MMA who require long-term maintenance of therapeutic concentrations, and it is expected to reduce the injection frequency from 2 to 3 times per week to once every 10–14 days.

3.6. Safety evaluation

MVLs are derived from naturally occurring lipids and have excellent biocompatibility and non-immunogenicity (Bapi et al., 2021). Therefore, the safety of the prepared MVLs was evaluated with emphasis on in vitro cytotoxicity, histopathology, hemocompatibility, hematological parameters, and serum biochemical parameters. The in vitro cytotoxicity of the MVL formulations was further assessed using 3T3 fibroblast cells via CCK-8 assay. As shown in Supplementary Fig. S8, after 24 h of incubation, both VB₁₂-MVLs and VB₁₂-β-MVLs maintained cell viability above 90% across all tested concentrations (0.02–1.2 mg mL⁻¹). These results indicate that neither formulation induces appreciable cytotoxicity in fibroblasts, further confirming the favorable local biocompatibility of the MVL platform. H&E staining results (Supplementary Fig. S9) showed no obvious pathological changes in the heart, liver, spleen, lungs, and kidneys of rats after administration compared with the control group, preliminarily indicating that the prepared VB₁₂-MVLs and VB₁₂-β-MVLs have low toxicity and good safety. Further evaluation of the injection site skin showed that the epidermis of the injection site skin in all four groups was relatively wrinkled, with no obvious hyperplasia of the spinous layer, intact hair follicles, no obvious glandular hyperplasia, no obvious inflammatory cell infiltration, and no obvious collagen fiber hyperplasia (Fig. 6A), demonstrating that the rats experienced minimal local irritation after treatment with VB₁₂-MVLs and VB₁₂-β-MVLs, and no adverse reactions were observed around the injection site. Next, an in vitro hemolysis assay of VB₁₂-MVLs was performed. The results showed that, compared with the positive control (1% Triton X-100) which caused complete hemolysis (Fig. 6B), VB₁₂-MVLs at concentrations as high as 4 mg mL⁻¹ showed no obvious hemolysis after incubation with 2% RBC suspension for 3 h, demonstrating good hemocompatibility of VB₁₂-MVLs. Finally, hematological and serum biochemical parameters were evaluated. The results for hematological parameters (white blood cell count, lymphocyte count, monocyte count, neutrophil count, and their percentages) were all within normal ranges (Fig. 6C-F and Supplementary Figs. S10A-S10C). The results for serum biochemical parameters (ALT, AST, CREA, and UREA) were also within normal ranges (Fig. 6G-H and Supplementary Figs. S10D-S10E). These results collectively demonstrate that the VB₁₂-MVL and VB₁₂-β-MVL formulations have good biocompatibility.

Fig. 6.

Safety evaluation of VB₁₂-MVLs and VB₁₂-β-MVLs. (A) Pathological results of skin tissue at the administration site of rats. (B) Hemolysis of different preparations after incubation for different times (1: 0.9% saline, 2: 2% Triton-X100, 3: 1 mg mL⁻¹ VB₁₂-MVLs, 4: 2 mg mL⁻¹ VB₁₂-MVLs, 5: 4 mg mL⁻¹ VB₁₂-MVLs). (C-H) The results of hematological indexes and blood biochemical examination of rats after administration. The range between the two dashed lines represents the normal reference range. ns: (P > 0.05), *P < 0.05, **P < 0.01, and ***P < 0.001.

4. Discussion

This study successfully developed a long-acting VB₁₂ delivery platform based on MVL technology and demonstrated that the incorporation of DM-β-CD significantly enhances both drug loading and sustained-release performance. The optimized VB₁₂-MVLs and VB₁₂-β-MVLs exhibited high encapsulation efficiency, favorable particle size, and a characteristic multi-chambered “honeycomb” structure. The introduction of DM-β-CD not only nearly doubled the drug concentration but also further extended drug release, likely due to the formation of inclusion complexes that retard diffusion. The in vivo depot effect was confirmed by fluorescence imaging, and the pharmacokinetic data demonstrated a marked prolongation of MRT (up to 121 h) and a 1.9-fold increase in relative bioavailability compared with the non-cyclodextrin formulation.

Compared with previously reported VB₁₂ delivery systems, our MVL-based approach offers a unique combination of high loading, sustained release, and biocompatibility. Conventional VB₁₂ formulations—whether free injection or oral preparations—are limited by rapid clearance and poor bioavailability (Butler et al., 2006; Fan et al., 2024; Liu et al., 2025; Martínez-Navarrete et al., 2026; Watanabe, 2007; Yu et al., 2021; Zhou, 2023). While some efforts have explored liposomal or nanoparticle-based VB₁₂ carriers, most have focused on unilamellar or multilamellar liposomes, which suffer from the “all-or-nothing” release problem and inferior mechanical stability (Chalasani et al., 2007a; Chalasani et al., 2007b). The multivesicular architecture of MVLs addresses these limitations by providing multiple non-concentric aqueous chambers that enable gradual, erosion-dependent drug release rather than burst release upon vesicle rupture. This structural advantage, combined with the high internal aqueous phase volume fraction, makes MVLs particularly suitable for hydrophilic drugs like VB₁₂.

The use of microfluidic technology in our preparation ensures scalable and reproducible production, addressing a key translational hurdle for liposomal formulations. The improved microfluidic-double emulsification method yielded particles with uniform size distribution and consistent batch-to-batch performance, which is essential for clinical translation.

The finding that DM-β-CD not only increased drug loading but also further prolonged release warrants mechanistic consideration. Importantly, the role of DM-β-CD was not to enhance EE—which was already high with the MVLs structure—but to overcome the drug loading bottleneck by increasing the apparent solubility of VB₁₂ in the internal aqueous phase. This distinction is critical: while EE reflects the proportion of drug successfully encapsulated relative to the total amount added, drug loading (i.e., the absolute drug concentration in the formulation) determines the practical utility of the formulation for clinical application. By elevating the drug concentration from 10.96 to 17.60 mg mL⁻¹, DM-β-CD enables a ∼60% reduction in injection volume for the same therapeutic dose, which is particularly advantageous for pediatric patients with MMA who require long-term treatment. Additionally, the sustained dissociation of the VB₁₂-β-CD complex within the aqueous chambers adds an extra kinetic barrier to drug diffusion, further contributing to the prolonged release profile. Molecular docking simulations suggested that VB₁₂ adopts an “up-and-down” binding mode with DM-β-CD without entering the cavity, primarily mediated by hydrogen bonding and other weak interactions (Jiang et al., 2011). This mode of interaction differs from classical inclusion complexation and may explain why the complexation did not disrupt the MVLs architecture (as confirmed by TEM and CLSM) while still achieving enhanced solubilization. The absence of chemical interaction between the drug and excipients, as confirmed by FTIR, further supports the physical nature of this enhancement.

The delayed release observed with VB₁₂-β-MVLs relative to VB₁₂-MVLs may be attributed to the sustained dissociation of the VB₁₂-β-CD complex within the aqueous chambers, which adds an additional kinetic barrier to drug diffusion across the lipid bilayers. This dual-level sustained release—from both the MVLs structure and the cyclodextrin complex—represents a novel strategy for prolonging the action of hydrophilic small-molecule drugs.

These findings are particularly relevant for pediatric MMA patients, who require long-term maintenance of therapeutic VB₁₂ levels. The extended release profile could reduce injection frequency from 2 to 3 times per week to once every 10–14 days, thereby improving adherence and quality of life. This is especially significant for children, for whom frequent injections pose substantial physical pain and psychological stress (Castelli et al., 2011).

Moreover, the excellent safety profile—no hemolysis, minimal local irritation, and normal hematological/biochemical indices—supports the translational potential of this platform. The use of endogenous lipids and the biocompatible cyclodextrin derivative further minimizes the risk of adverse reactions, making this formulation particularly attractive for chronic use in pediatric populations.

Nevertheless, some limitations exist. First, the long-term stability of the lyophilized formulation requires further optimization. While sodium alginate showed promising protective effects, the EE after reconstitution (66.71%) still leaves room for improvement. Second, the pharmacokinetic study was performed in healthy rats rather than MMA disease models. Disease-associated metabolic alterations might affect the pharmacokinetic profile and should be evaluated in future studies. Third, chronic toxicity and efficacy studies in relevant animal models are needed before clinical translation. Additionally, while our preliminary safety evaluation in rats was reassuring, comprehensive toxicological assessment following regulatory guidelines will be necessary for investigational new drug application.

Future work could explore alternative routes of administration (e.g., intraperitoneal or intravenous) or combination therapies with other metabolic modulators. The MVL platform itself is versatile and could be adapted for other hydrophilic drugs requiring long-acting delivery, expanding its therapeutic impact beyond MMA.

Despite these limitations, our results provide a strong proof-of-concept for a long-acting VB₁₂ formulation. The successful integration of MVL technology, microfluidic manufacturing, and cyclodextrin-mediated solubilization offers a feasible paradigm for developing hydrophilic small-molecule drugs into long-acting depot formulations.

5. Conclusions

In this study, a long-acting VB₁₂ delivery platform based on MVLs technology (VB₁₂-MVLs) was successfully developed, and a high-drug-loading formulation (VB₁₂-β-MVLs) was constructed by innovatively introducing DM-β-CD. Both formulations exhibited uniform particle size, high EE, a typical multi-compartment “honeycomb-like” structure, and good in vitro stability. Notably, VB₁₂-β-MVLs not only nearly doubled the formulation drug concentration but also achieved more sustained drug release. In vivo studies confirmed that this platform forms a subcutaneous drug depot after injection, significantly prolonging the drug residence time in vivo (MRT extended to 121 h) and greatly improving bioavailability (up to 191.89%). Preliminary safety evaluation results were satisfactory. In conclusion, the developed VB₁₂-MVLs, especially VB₁₂-β-MVLs, as a safe and long-acting new formulation, have great translational potential for improving the treatment of methylmalonic acidemia and other diseases requiring long-term VB₁₂ supplementation. This strategy also provides a feasible paradigm for developing hydrophilic small-molecule drugs into long-acting depot formulations.

CRediT authorship contribution statement

Rui Chang: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Funding acquisition, Formal analysis. Yu Lu: Writing – original draft, Visualization, Validation, Investigation, Formal analysis. Yiming Liu: Validation, Investigation. Qian Gao: Validation, Investigation. Xuewei Zhang: Validation, Investigation. Yanan Zhang: Validation, Investigation. Yutong Xing: Validation, Investigation. Xue Yan: Writing – original draft. Chaoxing He: Validation, Supervision, Conceptualization. Huifeng Zhang: Supervision, Funding acquisition, Conceptualization. Bai Xiang: Writing – review & editing, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.

Ethics declaration

This study was conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines.

This study was approved by the Laboratory Animal Ethical and Welfare Committee of Hebei Medical University.

(Approval No. IACUC-Hebmu-2024057)

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material: Supplementary Information for Fabrication of vitamin B₁₂-loaded multivesicular liposomes via microfluidic double-emulsification: Extended pharmacokinetics and bioavailability

Data availability

Data will be made available on request.