Charge/stiffness-tunable nano-microsphere overcomes intestinal mucosal barrier for rheumatoid arthritis treatment

TL;DR

A tunable nano-microsphere system was developed to overcome intestinal barriers for more effective rheumatoid arthritis treatment.

Contribution

A dynamically tunable nano-microsphere system with charge and stiffness adaptability for intestinal drug delivery is introduced.

Findings

The nano-microspheres achieved a 1.6-fold increase in maximum blood concentration compared to conventional carriers.

MLVs demonstrated effective penetration of intestinal barriers and multimechanistic endocytosis potential.

DAPC in the vesicle shell enables charge reversal via non-classical hydrolysis, confirmed by LC-MS/MS analysis.

Abstract

Intestinal drug delivery is a crucial route for rheumatoid arthritis (RA) therapy. However, its effectiveness is often hampered by the viscosity gradient of the mucus layer and the selective degradation and efflux functions of the epithelial barrier. To address these challenges, we developed a nano-microsphere system featuring charge- and stiffness-tunable multilayered vesicles (MLVs) encapsulated within pH-sensitive microspheres. The MLVs are engineered to traverse the negatively charged, viscosity-gradient mucus by sequentially shedding their flexible shells and undergoing a positive-to-negative charge reversal. This exposes a rigid, neutral core that enables multimechanistic endocytosis with potential for transcellular transport. The dynamic tunability of the MLVs is attributed to the incorporation of di-artesunate-phosphatidylcholine (DAPC) in the vesicle shell. LC-MS/MS analysis reveals that DAPC undergoes terminal hydrophobic chain carboxylation via a non-classical hydrolysis pathway, facilitating the observed charge reversal. Encapsulation within pH-sensitive microspheres further protects the MLVs from premature degradation and ensures targeted intestinal delivery. Cryo-electron microscopy and in vitro studies confirmed the multilayered architecture, dynamic adaptability, and effective penetration of intestinal barriers by the MLVs and the nano-microspheres. In vivo, this system achieved a 1.6-fold increase in maximum blood concentration compared to conventional carriers, alongside significantly enhanced therapeutic efficacy for RA. In summary, we present a dynamically adaptive, intestinal barrier-penetrating nano-microsphere platform that offers a promising strategy for RA treatment.

Figures

F–I, S19](#fig4), Table S2) showed that MLV@MS most effectively reduced pro-inflammatory cytokine expression, followed by CV@MS, with no significant difference between MS and control. This effect is attributed to improved MTX delivery.

Full text

Introduction

1

Rheumatoid arthritis (RA) is a systemic autoimmune disease characterized by chronic synovitis[1]. Disease-modifying antirheumatic drugs (DMARDs), such as methotrexate (MTX), remain the cornerstone of oral RA therapy [1,2]. However, the clinical efficacy of DMARDs is often compromised by poor intestinal absorption, limited bioavailability, and significant side effects [[3], [4], [5]]. Nanomedicines, especially lipid-based carriers like liposomes, offer the potential to enhance drug stability and bioavailability while mitigating side effects [6]. Despite these advantages, conventional nanomedicines face significant challenges in traversing the negatively charged, viscosity-gradient mucus layer and achieving selective epithelial transport within the intestine, as these barriers require distinct physicochemical properties in terms of surface charge and mechanical stiffness [[7], [8], [9]]. Therefore, designing nanomedicines capable of dynamically regulating both charge and stiffness to adapt to the unique demands of each transport stage has become a critical strategy for overcoming the barriers to oral DMARD delivery.

During intestinal transport, nanomedicines must overcome two primary barriers: the mucus layer and the intestinal epithelium, each with specific requirements for charge and rigidity [10]. The mucus layer, serving as the first line of defense, consists of a negatively charged mucin network exhibiting a gradient from high viscoelasticity to dense compaction [11]. Moderately positive surface charges enhance adhesion and prolong retention within the mucus, promoting subsequent transport. However, excessive positive charge may result in over-adhesion and hinder penetration into deeper, denser mucus layers [12]. Thus, an optimal, finely-tuned positive charge is necessary to minimize electrostatic repulsion while enabling effective mucus penetration. Furthermore, a flexible membrane with high fluidity allows vesicles to more readily navigate the mucus network, enhancing penetration efficiency [13,14]. In contrast, during epithelial transcytosis, the requirements shift. The epithelial cell membrane carries a weak negative charge; moderate positive charges facilitate nanoparticle binding via electrostatic attraction and promote endocytic uptake. However, to avoid cytotoxicity and non-specific uptake, nanoparticles should possess a mildly positive or near-neutral surface charge [14,15]. Stiffness also plays a critical role: stiffer nanoparticles exhibit enhanced structural stability, evade lysosomal degradation, and are more likely to undergo macropinocytosis or paracellular transport, thereby improving drug protection and delivery efficiency [16]. Moreover, stiffer particles are more likely to undergo macropinocytosis or paracellular transport, enabling better drug protection and efficient delivery [16]. Consequently, the ability to dynamically regulate both charge and stiffness is essential for overcoming the sequential challenges posed by the mucus and epithelial barriers, ensuring efficient intestinal transport and therapeutic efficacy.

Liposomes and related vesicles inherently offer advantages for modulating charge and stiffness due to the flexibility of their lipid bilayers. For instance, introducing charged phospholipids (e.g., positively charged DOTAP or negatively charged phosphatidylglycerol) enables precise tuning of surface charge properties [17,18]. Modifying cholesterol content or selecting phospholipids with distinct phase transition temperatures (Tm) allows effective regulation of membrane fluidity and rigidity [19]. However, conventional vesicles are typically homogeneous in composition and lack the capacity for dynamic in vivo modulation of charge and stiffness, limiting their performance in navigating the complex intestinal environment. While microenvironment-responsive charge-reversal strategies can partially address this, achieving coordinated, dynamic regulation of both surface charge and stiffness during intestinal transit remains challenging [20,21]. To address these challenges, a promising strategy is the design of core–shell nanostructures with mismatched charge and stiffness, combined with the selective removal of the outer shell to enable dynamic adjustment. Specifically, a highly flexible, positively charged outer shell facilitates mucus penetration, while a neutral/negative-charged, rigid core promotes efficient epithelial transcytosis. Nevertheless, liposomes with low rigidity are often susceptible to instability or spontaneous fusion due to high membrane fluidity [22]. While introducing positive charges can alleviate fusion via electrostatic repulsion, further optimization is needed to enhance overall stability and performance. Previous studies have demonstrated that encapsulating nanoparticles in hydrogel microspheres markedly improves their stability [6,23,24]. Embedding vesicles in pH-responsive hydrogels, such as calcium alginate, not only protects against gastric acid and digestive enzymes but also enables targeted release in the neutral or weakly alkaline environments of the small intestine or colon, further improving delivery efficiency [18,25]. Therefore, integrating core–shell multilayer vesicles with calcium alginate microsphere encapsulation offers a promising approach for overcoming the intestinal transport challenges of nanomedicines via dynamic charge and stiffness regulation.

In previous studies, a di-artesunate-phosphatidylcholine (DAPC) was identified for its superior self-assembly characteristics and faster hydrolysis rate compared to traditional phospholipids, attributed to its conversion into active monomers under physiological conditions [6,18,26]. In this study, we further investigated the hydrolysis mechanism of DAPC and found that, in simulated intestinal fluid, it exhibited a different hydrolysis compared to the conventional sn-1/sn-2 hydrolysis of traditional phospholipids (Scheme 1, Fig. S1). This process exposed two carboxyl groups on the lipid chains, endowing DAPC with a unique charge-reversal capability. Leveraging its excellent self-assembly properties, DAPC was utilized as the responsive shell material for multilayer vesicles (MLVs). Through double-emulsion and self-assembly techniques, MLVs with tunable charge and stiffness were constructed and subsequently encapsulated in hydrogel microspheres using gas microfluidics, yielding a nano-microsphere platform for efficient intestinal delivery of oral DMARDs. Specifically, a double-emulsion extrusion method was used to optimize saturated phospholipid DSPC and high-cholesterol formulations to generate a neutral/negative-charged, high-stiffness HAP-1-modified liposomal core. Under low-shear conditions, DAPC, TAT-phospholipid, and a small amount of DOTAP were combined to assemble the flexible, responsive MLV shell, which was further TAT-modified. Finally, MLVs were encapsulated in calcium alginate microspheres via gas microfluidic technology to enhance stability during gastrointestinal transit and enable targeted intestinal release. Cryo-electron microscopy (Cryo-EM) and dynamic light scattering (DLS) confirmed the structure and tunable charge/stiffness features of the MLVs. In vitro studies demonstrated the nano-microsphere's ability to penetrate intestinal mucus and modulate synovial fibroblasts and macrophages. In vivo experiments using an adjuvant-induced arthritis (AIA) model showed that the charge- and stiffness-tunable nano-microsphere achieved efficient intestinal transport and delivered significant therapeutic effects in RA treatment. This study presents a novel strategy for transcellular intestinal transport of nanomedicines and provides a new technological platform for the oral treatment of chronic diseases.Scheme 1Charge/stiffness-tunable nano-microsphere to enhance intestinal transport for rheumatoid arthritis treatment: (top) Schematic diagram of high-rigidity, neutral-charge, small unilamellar vesicles (SUVs) loaded with MTX, prepared using the double emulsion method (W/O/W); (middle) Schematic diagram of multilamellar vesicles (MLVs) constructed with flexible, positively charged shells via self-assembly, enabling the generation of negative charges in a simulated intestinal environment through the non-classical hydration of DAPC, achieving charge and stiffness tuning; (bottom) Schematic diagram of calcium alginate hydrogel encapsulation constructed via gas-shear microfluidics, enabling targeted activation of MLVs in the intestine after oral administration. The non-classical hydration of DAPC enables charge and stiffness tuning, matching the charge and stiffness requirements of nanoparticles during intestinal transport.Scheme 1

Results and discussion

2

LC-MS/MS analysis of DAPC to detect hydrolysis mechanism

2.1

In previous studies, DAPC was synthesized through a heterogeneous esterification reaction between artesunate and glycerophosphatidylcholine (Fig. 1A) [27,28]. DAPC has been shown to possess liposome-forming properties similar to traditional phospholipids, while exhibiting significantly enhanced self-assembly activity [26]. Notably, its degradation behavior differs markedly from that of conventional phospholipids or other phospholipid conjugates [6]. In this study, we further investigated the degradation profile of DAPC in artificial intestinal fluid (AIF) using LC-MS/MS. Standard samples of DHA (the active intermediate of DAPC), artesunate (the parent drug), and DAPC were first analyzed to record their elution profiles and corresponding m/z values (Fig. 1B). Upon incubating DAPC with AIF, LC-MS/MS was employed to monitor the elution curves for the detectable m/z values of DHA, artesunate, and DAPC, and concentrations were quantified based on peak areas at their respective retention times. DHA and artesunate were used as controls. As shown in Fig. S2, after incubation for varying durations, the DHA peak area decreased moderately but retained about 50% of its initial level after 24 h, and artesunate showed a similar trend (Fig. 1C). In contrast, the DAPC peak area (m/z 990.4458) declined rapidly and was undetectable after 24 h (Fig. 1D). Importantly, the MRM mode was used to directly monitor the elution profile at the specific m/z (990.4458), allowing clear observation of DAPC degradation without interference from extraneous peaks. The minor peak observed at 8.3 min in the DAPC elution profile represents a fragment or isomer of DAPC rather than an impurity, as it shares the same molecular weight. These results indicate that DAPC undergoes significantly accelerated degradation compared to DHA and artesunate. It remains unclear whether this degradation is due to the ester bonds at the sn-1 and sn-2 positions, similar to traditional phospholipids, or derived from the ester bond within the artesunate structure. Additional HPLC and LC-MS analyses confirmed the purity of DAPC (Figs. S3–5).Fig. 1Synthesis and hydrolysis (non-sn-1/sn-2) mechanism of di-artesunate phosphatidylcholine conjugate (DAPC): (A) Synthetic pathway of DAPC; (B) m/z and retention time (RT) of different compounds detected by LC-MS/MS; (C) Elution curve of the molecular weight corresponding to artesunate after incubation of artesunate in artificial intestinal fluid (AIF) for a certain period; (D) Elution curve of the molecular weight corresponding to DAPC after incubation of DAPC in AIF for a certain period; (E) Elution curve of the molecular weight corresponding to artesunate after incubation of DAPC in AIF for a certain period; (F) Elution curve of the molecular weight corresponding to DHA after incubation of DAPC in AIF for a certain period; (G) Hypothetical hydrolysis mechanism of DAPC in AIF and the generation of negative charges.Fig. 1

To further clarify the degradation pathway, we analyzed the DAPC incubation solution by Extracted Ion Chromatogram (EIC) mode, focusing on the signals corresponding to artesunate and DHA (Fig. 1E and F). The results showed that artesunate was almost undetectable, while the integrated DHA signal area corresponded closely to the decrease in DAPC. Since artesunate itself is relatively stable, we speculate that DAPC degradation primarily results from hydrolysis of the ester bond within the artesunate structure, rather than classical hydrolysis at the sn-1 or sn-2 positions of the phospholipid backbone. This atypical hydrolysis may be due to the high hydrophilicity and unique amphiphilic structure of DAPC, which accelerates hydrolysis of the conjugated artesunate ester bond. Nevertheless, the specific mechanism warrants further investigation.

Although sPLA1/2 catalyze sn-1/sn-2 hydrolysis, the in vitro degradation medium for DAPC contained no sPLA enzymes. Nevertheless, we observed accelerated ester hydrolysis that is absent in conventional phospholipids (e.g., DSPC), which we term “non-classical hydrolysis” (Fig. 1G). Similar behavior has been reported previously, including the accelerated release of dihydroartemisinin from DAPC and related, albeit slower, hydrolysis in other dimeric drug-phospholipid conjugates [26,29]. If such hydrolysis occurs on the fatty acid chains, it would result in terminal carboxylation. Under the neutral pH of the small intestine, this carboxylation would increase the negative charge, and the resulting drastic chemical changes in the phospholipid could cause its assemblies to loosen or disintegrate. We also speculate that the significant steric hindrance of DHA may render its ester bond more susceptible to hydrolysis under certain conditions. Therefore, introducing other sterically hindered alcohols, such as triphenylmethanol, 1-adamantanol, or tert-butanol, to synthesize phosphatidylcholine derivatives may also yield similar hydrolytic effects.

DAPC can be used in conjunction with DOTAP to prepare MLVs with tunable charge from positive to negative. Since the adhesion of positively charged nanocarriers to the intestinal mucosa occurs relatively rapidly [30], this hydrolysis rate difference can be leveraged to achieve rapid charge reversal and dynamic changes in assembly structure during mucus layer penetration (Scheme S1, Figure S1).

Preparation of charge/stiffness-tunable MLVs with DAPC

2.2

Based on the unique properties of DAPC, including its rapid hydrolysis-induced charge reversal and structural changes, we incorporated it into the shell of MLVs to achieve tunable charge and rigidity, thus meeting the dynamic requirements for intestinal transport.

We constructed MLVs using a water-in-oil-in-water (W/O/W) double emulsion method combined with the electrostatically induced self-assembly of flexible phospholipids, achieving MTX loading and the preparation of a “onion-like” structure with stiffness/charge heterogeneity. Here, the double emulsion method used is very similar to traditional vesicle preparation methods, but more extensive film homogenization was employed to construct small unilamellar vesicles (SUV) as much as possible [31]. To maximize the solubilization of MTX, a sodium bicarbonate solution was initially selected as the inner aqueous phase. As the outer aqueous phase was added, the pH was adjusted to neutral, causing MTX to precipitate, which also facilitated the formation of a rigid solid-phase core [32].

After obtaining SUVs, additional phospholipid components, including positively charged DOTAP, DAPC, and weakly positively charged DSPE-PEG-TAT, were assembled onto the SUV core through electrostatic and hydrophobic interactions under slow shear conditions [33]. DAPC has already been demonstrated to possess self-assembly capabilities surpassing those of traditional phospholipids [26]. Together with DOTAP and DSPE-PEG-TAT, under the combined guidance of hydrophobic interactions and electrostatic forces, it gradually forms the unique MLVs structure of phospholipids. After ultrafiltration to remove organic solvents, the MLVs were successfully formed.

As shown in Fig. 2A, Cryo-EM images sequentially display blank SUVs, MTX-loaded SUVs, and MTX-loaded MLVs. Notably, MTX loading markedly increases the contrast of the SUV inner aqueous phase, likely due to electron absorption by MTX. The characteristic onion-like multilamellar structure of the MLVs is clearly visible, and vesicle size increases accordingly, with SUVs ranging from 50 to 200 nm and MLVs averaging around 200 nm. This multilayered structure is central to the tunable charge and stiffness properties of MLVs, as layers of flexible, positively charged shells encapsulate and protect the SUV core from premature exposure.Fig. 2Preparation of MLV@MS nano-microsphere: (A) Cryo-EM images of Blank SUV, MTX-loaded SUV, and MTX-loaded MLV, scale bar = 50 nm; (B-D) Particle size distribution of (B) H-MLV, (C) L-MLV, and (D) MLV over time; (E) Zeta potential changes of SUV and MLV over time in PBS (n = 3); (F) SEM images of MLV@MS; (G) Morphology of Blank MS and MLV@MS under an optical microscope, scale bar = 100 μm; (H) Particle size information of Blank MS and MLV@MS; (I) MLV@MS in three simulated digestive fluids (AGF: artificial gastric fluid; AIF: artificial intestinal fluid; ACF: artificial colonic fluid), scale bar = 100 μm (n = 3); (J) MTX release from MTX-loaded MLV@MS in simulated digestive fluids (n = 3).Fig. 2

To compare the stiffness and fluidity of different phospholipid architectures, we prepared H-MLV and L-MLV using only the inner- or outer-layer phospholipid components, respectively, following the same protocol as for MLVs. Stiffness characteristics were assessed by monitoring size distribution changes over time via dynamic light scattering (DLS) [34]. As shown in Fig. 2B–D, vesicles composed of low-stiffness phospholipids exhibited the greatest particle size variation and significantly broadened size distributions after 2 h, while high-stiffness MLVs maintained nearly constant particle size in the same period. MLVs with heterogeneous stiffness/charge maintained their size distribution well for up to 1 h, with gradual broadening over time. Moreover, atomic force microscopy (AFM)-derived modulus curves are shown in Fig. S6, and the corresponding Young's moduli are summarized in Fig. S7. These data directly quantify vesicle stiffness and are consistent with the prior DLS results. Notably, there is a clear stiffness contrast between the MLV shell and the internal SUV core. As the lower-modulus shell is shed, the overall modulus of the vesicle increases substantially. These results suggest that vesicles formed from low-stiffness phospholipids, even with charge repulsion, are prone to fusion and structural instability, leading to reduced particle size uniformity. Although MLVs may experience similar issues, they retain structural stability in the short term. As DAPC hydrolyzes and negative charge increases, the flexible outer layer of the MLV may deform or detach, gradually exposing the rigid, negative-charged SUV core beneath.

To investigate this process, we measured the zeta potential of MLVs after incubation in PBS buffer, using Sephadex G-50 column separation. As shown in Fig. 2E, the initially high positive charge (∼20 mV) of MLVs decreased over time, stabilizing near neutral (<+5 mV) after 3∼4 h. In contrast, SUVs exhibited a zeta potential of approximately −15 mV (Fig. S8). Similar trends were observed for both MLV and SUV zeta potentials in AIF (Fig. S9).

These data demonstrate that the positive and negative charges of MLVs ultimately reach equilibrium. This dynamic instability in charge and stiffness is precisely what enables MLVs to traverse the intestinal barrier: the initial positive charge and flexible surface facilitate rapid penetration through the mucus layer, while the subsequent high stiffness and neutral or mildly positive surface promote efficient epithelial cell uptake via endocytosis. However, these findings also suggest that MLVs alone are unlikely to remain structurally stable during storage and oral delivery, highlighting the necessity for additional hydrogel encapsulation to ensure stability.

In this work, both MLVs and L-MLVs differ fundamentally from conventional vesicles. L-MLVs are a control with unsaturated outer shells instead of DSPC, making them intrinsically unstable. Our MLVs use high phase transition DSPC mainly in the inner core, while the outer shell has a low phase transition, low cholesterol composition, yielding a very low overall Young's modulus by design to enable rapid shell shedding during delivery [35]. Although such low modulus architectures are typically unstable, we stabilize them by encapsulating the system in calcium alginate microspheres, effectively freezing the structure. These combined features are absent in conventional formulations, supporting our use of very low modulus MLVs and outer shells in this study.

Preparation of nano-microsphere with calcium alginate microsphere encapsulation

2.3

Physical encapsulation of vesicles within hydrogels to form nano-microspheres has been widely reported to enhance vesicle stability [24]. In our system, the relatively unstable structure of charge/stiffness-tunable MLVs particularly benefits from this encapsulation, which preserves their integrity upon exposure to the intestinal environment, effectively “freezing” them in place. Notably, this encapsulation process does not damage the vesicles, including the onion-like multilayered architecture of MLVs. Upon hydrogel disintegration, MLVs need only shed their outermost positively charged layer to separate from the alginate matrix.

We employed a classic gas microfluidic method, as previously described [25], enabling highly efficient, large-scale production of microspheres that instantly gel upon contact with calcium chloride solution, forming MLV@MS nano-microspheres. This approach effectively preserves MLV structure. As shown in Fig. 2F, SEM images reveal that MLV@MS maintain uniform spherical morphology. The enlarged surface displays highly regular wrinkling, likely due to intrinsic properties of calcium alginate, while multiple depressions observed on the sphere surface likely form during hydrogel drying, as they are absent under optical microscopy (Fig. 2G). Both Blank MS and MLV@MS appear as smooth spheres under light microscopy; Blank MS is more transparent, whereas MLV@MS appears turbid, likely due to the encapsulated MLVs, which exist as a suspension. There is minimal difference in particle size between Blank MS and MLV@MS, as confirmed by particle size distribution analysis (Fig. 2H). Drug loading (DL) and encapsulation efficiency (EE) of MTX in nanomicelles and microspheres were calculated and shown in Table S1.

Hydrogel microsphere encapsulation maintains the stability of charge/stiffness-tunable MLVs in storage solutions (high-Ca^2+^ or low-pH) [25]. Replacing calcium chloride with soda water still yielded well-formed MLV@MS, enhancing the potential acceptability of oral microspheres for patients.

To further characterize MLV@MS responsiveness in gastrointestinal environments, we incubated MLV@MS with artificial gastric fluid (AGF), intestinal fluid (AIF), and colon fluid (ACF) [36]. The structure of MLV@MS was observed under an optical microscope, as shown in Fig. 2I. Both ACF and AIF rapidly disrupted MLV@MS, with ACF producing the fastest effect: most microspheres disappeared within 0.5 h, and almost none remained after 1 h. AIF caused most microspheres to disappear after 1 h and nearly all were gone by 2 h. In contrast, AGF preserved microsphere integrity even after 6 h, with only minor shape irregularities. Since MLV@MS rapidly transits the stomach during oral administration, the gastric environment has negligible effect. However, rapid disintegration occurs in the small intestine and colon, releasing MLVs at these sites. This in vitro degradation primarily reflects hydrogel shell disintegration rather than the breakdown of the loaded MLV nanocarriers themselves. Our delivery system is intentionally designed so that rapid hydrogel microsphere degradation triggers timely release of MLVs in the small intestine. Once released, MLVs with tailored surface charges can rapidly interact with the intestinal mucosa, ensuring efficient mucoadhesion and drug delivery. The key is not prolonged retention in the intestinal lumen, but prompt and precise localization and mucosal penetration.

Disintegration of the microsphere structure exposes the encapsulated charge/stiffness-tunable MLVs, allowing rapid shedding of the MLV's outermost phospholipid layer and release into the environment. MTX release kinetics were characterized by HPLC measurement of MTX in the supernatant after low-speed centrifugation (Fig. 2J). ACF induced the fastest MTX release, nearly complete within 10 min, while AIF induced a slightly slower release, up to 1 h. Given that MLV@MS first traverses the small intestine after oral administration, this sustained release enables uniform MLV@MS distribution and release along a larger intestinal segment, with the remainder rapidly disintegrating in the colon. This release profile facilitates efficient absorption throughout the intestinal tract. It is likely that the detected MTX is predominantly in the MLV-loaded form, as free MTX is poorly soluble in neutral aqueous phases.

In addition, we evaluated the storage stability of MLV@MS under three scenarios. First, we assessed particle size changes after lyophilization followed by rehydration and swelling, measured under an optical microscope. This reflects the practical storage form, as MLV@MS is stored long-term in a lyophilized state, with calcium alginate serving as an effective lyoprotectant for MLVs. Second, we monitored particle size changes in storage buffer at 4 °C and 25 °C to evaluate stability at low and ambient temperatures, respectively. All results (Fig. S10) indicate that MLV@MS remains stable under these different storage conditions.

These results strongly confirm that calcium alginate microspheres can effectively encapsulate MLVs, maintaining their integrity during oral administration and ensuring efficient release in the intestinal system.

In vitro cytotoxicity assay and cellular uptake study

2.4

The cytotoxicity of MLV@MS and the uptake behavior of released MLVs by intestinal epithelial cells were further evaluated. To ensure a comprehensive assessment, blank microspheres (MS) were prepared as a control, and a controlled vesicle nano-microsphere (CV@MS) was generated using the same method as MLV@MS, except DAPC was replaced with DSPC and an equimolar amount of artesunate.

Cytotoxicity was assessed using standard cell live/dead staining and CCK-8 assays. For live/dead staining, Caco-2 cells were incubated with blank MS, CV@MS, and MLV@MS (final concentration: 10 μg/mL) for 24 and 48 h. After staining with Calcein-AM and propidium iodide (PI), the cells were observed under a fluorescence microscope. As shown in Fig. 3A, nearly all Caco-2 cells remained viable across all groups, with negligible differences. Furthermore, cell numbers increased significantly after 48 h, and growth trends were nearly identical among groups. At the same microsphere concentration, cell proliferation was further assessed by CCK-8 assay. As shown in Fig. 3B, Caco-2 cells exhibited continuous proliferation over a 5-day period, with minimal differences between the groups and no statistically significant differences observed (P > 0.05). Similarly, we performed the same CCK-8 assay using a commonly used mouse fibroblast cell line, L929. As shown in Fig. 3C, the proliferation characteristics of L929 cells were comparable to those of Caco-2 cells. These findings from both cell lines indicate that MLV@MS does not exhibit significant cytotoxicity.Fig. 3In vitro cytotoxicity and cellular uptake assays: (A) Live/dead staining of Caco-2 cells after incubation with MLV@MS, scale bar = 100 μm (n = 3); (B) Cell proliferation of Caco-2 cells and (C) L929 cells after incubation with MLV@MS (n = 3); (D) and (E) Flow cytometry data and semi-quantitative analysis of DiO-labeled MLV uptake by Caco-2 cells (n = 3); (F) and (G) Laser confocal microscopy imaging and semi-quantitative analysis of DiO-labeled MLV@MS uptake by Caco-2 cells, scale bar = 200 μm. (n = 3) (n.s.: no significance, ***P < 0.001).Fig. 3

To compare vesicle uptake efficiency by Caco-2 cells, flow cytometry was used. CV (SUV) and MLV were labeled with DiO, and an equivalent DiO amount in PBS served as a control. As shown in Fig. 3D, both MLV and CV exhibited a significantly higher level of uptake by Caco-2 cells compared to the Blank Control and DiO Control, with uptake quantities showing orders of magnitude improvement. Notably, MLV uptake was significantly higher than CV uptake (P < 0.001), as shown quantitatively in Fig. 3E. We speculate that the enhanced cellular uptake of MLVs primarily stems from the introduction of positively charged components, which strengthen electrostatic interactions with the negatively charged cell membrane.

To evaluate cellular uptake following microsphere encapsulation, CLSM was used to compare DiO uptake after incubation with CV@MS and MLV@MS. The DiO supernatant was used as the control. As shown in Fig. 3F and G, confocal images revealed that both CV@MS and MLV@MS substantially enhanced cellular uptake compared to the control, with MLV@MS exhibiting a stronger intracellular fluorescence signal, indicating higher uptake efficiency. In view of the differing endocytic pathways underlying transcytosis and lysosomal trafficking, we performed additional experiments with a panel of endocytic inhibitors to quantify each pathway's contribution to MLV internalization, which clarified the dominant uptake mechanisms of MLVs. As shown in Fig. S11A, all chlorpromazine, filipin and dynasore demonstrated inhibition of cellular uptake. And dynasore preincubation led to the most pronounced reduction in cellular uptake in Caco-2 cells, compared with chlorpromazine and filipin. Inhibitor profiling indicates that multimechanistic endocytosis is involved including dynamin-dependent and cholesterol-rich raft/caveolae pathways, which are commonly associated with transcytosis, suggesting a potential for transcellular transport. As shown in Fig. S11B, cytochalasin D (CytoD), indicating that actin-dependent endocytosis plays a major role in their internalization [37].

Moreover, based on the available data, lysosomal involvement in MLV transport cannot be fully excluded. Our design, positively charged with a soft and rapidly shedding outer shell, is intended to promote membrane fusion and early endosomal escape, thereby reducing deep lysosomal routing, consistent with the observed uptake kinetics. Even if a fraction traffics to lysosomes, early release or escape can preserve MTX activity and support transcellular flux. In future work, we will conduct lysosomal co localization and inhibition assays to quantitatively assess this impact.

In summary, MLVs demonstrate significantly accelerated cellular uptake compared to CVs, a property retained even after microsphere encapsulation, suggesting that the encapsulation process preserves the unique structure and functionality of MLVs. The enhanced uptake is likely due to the relatively positive surface charge of MLVs, promoting stronger interactions with the negatively charged cell membrane and facilitating more efficient internalization.

Ex vivo evaluation of intestinal barrier transport efficiency

2.5

To more accurately assess the intestinal barrier-crossing capability of MLV@MS, we employed an ex vivo rat small intestine model [38]. Briefly, a 3 cm intestinal segment was excised, both ends ligated, and the lumen filled with a suspension of AIF and MLV@MS (Fig. 4A). DiO supernate and CV@MS served as controls. The segment was immersed in 5 mL of external medium and incubated at 37 °C for 1 h. After incubation, the segment was sectioned and stained with DAPI. The external medium was collected to incubate Caco-2 cells for 1 h, followed by DAPI staining and CLSM observation.Fig. 4In vitro transintestinal transport and anti-RA evaluation: (A) Schematic diagram of the in vitro transintestinal transport evaluation; (B) Immunofluorescence sections showing the distribution of DiO-labeled MLV in the small intestine, scale bar = 500 μm (n = 3); (C) Immunofluorescence images of the medium supernatant after incubation with Caco2 cells, scale bar = 100 μm (n = 3); (D) Scratch assay images of MH7A cells; (E) Immunofluorescence staining of iNOS and CD206 in RAW264.7 cells after incubation with MLV@MS, scale bar = 50 μm (n = 3); (F) and (G) mRNA expression levels of IL-1β and TNF-α in RAW264.7 cells after incubation with MLV@MS (n = 3); (H) and (I) Secretion levels of IL-1β and TNF-α in RAW264.7 cells after incubation with MLV@MS (n = 3). (**P < *0.05, ***P < *0.01, ***P < 0.001).Fig. 4

Staining results (Fig. 4B) showed evident DiO fluorescence in the submucosa of sections treated with MLV@MS, demonstrating effective mucosal penetration after microsphere release. The DiO fluorescence intensity for the MLV@MS group was significantly higher than for the other groups, indicating superior mucosal penetration compared to CV. When the external medium was incubated with Caco-2 cells (Fig. 4C), cells exposed to MLV@MS exhibited significantly greater DiO uptake than those treated with CV@MS, as further quantified in Fig. S12. This confirms that, after crossing the intestinal barrier, MLV@MS delivers its cargo efficiently to the basolateral side for cellular uptake, highlighting its superior translocation and delivery efficiency.

In addition, a visual CLSM model and Transwell model to assess the transmembrane transport of MLV according to a published work with some modification [39]. Firstly, we co-incubated MLVs with a simulated mucus system (low-concentration porcine mucin) to monitor morphological changes, and then loaded DiD to visualize mucus penetration via z-stack imaging and 3D reconstruction. These visual data provide a more intuitive assessment of the vesicles’ ability to traverse the mucus layer (Fig. S13). Then, we performed semi-quantitative fluorescence measurements in the basolateral chamber to evaluate the amount internalized by cells. After co-culturing Caco-2 and HT29-MTX cells in Transwell inserts for 21 days, we conducted MLV transport assays (Fig. S14A). FITC was used to label MLV (as a surrogate for MTX, which is also hydrophobic). Fluorescence from the mixed cell layer in the apical chamber and from the medium in the basal chamber (after treatment with 1% Triton X-100 for 5 min) was assessed by fluorescence microscopy and a fluorescence plate reader, respectively. As shown in Fig. S14B, the Caco-2 and HT29-MTX cell layer in the apical chamber exhibited residual FITC fluorescence for both MLV and CV under fluorescence microscopy. As shown in Fig. S14C, the fluorescence intensity of MLV in the basal chamber was markedly higher than that of the CV group (P < 0.001).

Meanwhile, we co-incubated MLVs with AIF and monitored the time-dependent changes in Young's modulus by AFM. As shown in Fig. S15, the initial modulus was approximately 0.5 MPa, increased to around 1.5 MPa after 10 min, and then rose more slowly to about 2 MPa at 1 h. For comparison, bare SUVs exhibited a modulus of approximately 3 MPa. Although the Young's modulus of MLVs remains relatively low within this range of variation, the stiffness changes are still discernible from the data. Meanwhile, this timescale is shorter than the actual intestinal penetration time, indicating that the triggering efficiency in AIF is well aligned with the transmembrane transport of MLVs. We speculate that multiple factors govern the shedding of the MLV outer shell, and that the simulated intestinal fluid is simplified relative to the in vivo environment, which may slow the detachment of the flexible shell and keep the Young's modulus within a relatively low range. Nevertheless, ongoing hydrolysis clearly and continuously modulates the modulus over time. Accordingly, we consider the AIF-triggered response to proceed progressively along the intestinal tract rather than in a single step.

The literature reports a small-intestinal residence/transit time of approximately 3∼4 h [40]. In this study, the stiffness of MLVs increased from ∼0.5 MPa to ∼2 MPa within about 1 h and then remained relatively stable for several hours. This indicates that MLVs complete the major charge and stiffness transition within roughly the first hour, with a stable plateau thereafter, which broadly matches the small-intestinal transit window. These data suggest that the kinetics of the mechanical transition align with the in vivo window for effective absorption and mucosal contact: during the first hour, the particles are softer, which may facilitate initial adhesion and mucus penetration; thereafter, the higher-stiffness plateau may help maintain structural stability and sustained contact under mucus and shear conditions. The transition may be faster under stronger proximal conditions, whereas distal segments or postprandial shifts in ionic strength and shear could introduce modest delays; nonetheless, there is generally sufficient time for the response to complete during transit.

In addition, MLVs continue to exhibit higher cellular uptake than CVs. Regarding the increased cellular uptake, many factors may contribute, including, but not limited to: the flexibility of MLVs, surface charge, particle size, cholesterol exposure, and the formation of a protein corona. In addition, different internalization pathways can exhibit different uptake efficiencies. Based on the observed outcomes, one plausible explanation is that, relative to CV, the most salient feature of MLVs is the introduction of positive charge; moreover, the combination of positive charge with a flexible and tunable shell may facilitate membrane fusion. Thus, the enhanced cellular uptake of MLVs likely arises from multiple synergistic factors.

Lastly, we would like to further clarify that, beyond the nonclassical hydrolysis of DAPC, multiple intestinal factors may affect outer shell shedding and surface charge evolution in MLVs. These factors can act synergistically or antagonistically, modulating both deshelling and charge reversal, and similar influences may occur in conventional liposomes as well as in MLVs. However, their impact is secondary to the intrinsic hydrolytic behavior of DAPC within MLVs. This distinctive property of DAPC sets it apart from other phospholipids and related assemblies.

These results further indicate that, compared with conventional CV, MLV possesses superior mucosal layer penetration.

In vitro anti-RA activity

2.6

The therapeutic efficacy of MLV@MS against RA was evaluated using two primary RA synovial cell types: synovial macrophages (RA-MLS, modeled by LPS-induced RAW264.7) and synovial fibroblasts (RA-FLS, modeled by MH7A) [41]. MH7A cells were incubated with MS, CV@MS, or MLV@MS (final concentration: 100 μg/mL); in the CV@MS and MLV@MS groups, the calculated MTX concentration was 200 ng/mL. Cell migration was assessed at 24, 48, and 72 h using a scratch assay (Fig. 4D andS16). Both MLV@MS and CV@MS significantly inhibited cell migration compared to controls, attributable to MTX loading; empty microspheres showed no inhibitory effect.

For inflammatory polarization, immunofluorescence, qPCR, and ELISA were performed [42,43]. RAW264.7 cells were incubated with MS, CV@MS, or MLV@MS for 24 h. Cells were stained for iNOS (M1 marker) and CD206 (M2 marker) [44,45]. As shown in Fig. 4E (quantified in Figs. S17–18), both MLV@MS and CV@MS promoted a shift from M1 to M2 polarization, with MLV@MS showing a slightly higher M2/M1 ratio, likely due to enhanced uptake. qPCR and ELISA analyses of IL-1β and TNF-α expression (Fig. 4F–I, S19, Table S2) showed that MLV@MS most effectively reduced pro-inflammatory cytokine expression, followed by CV@MS, with no significant difference between MS and control. This effect is attributed to improved MTX delivery.

Transcriptomic sequencing further confirmed the anti-inflammatory effects. Hierarchical clustering (Fig. 5A) and volcano plots (Fig. 5B–D) revealed that LPS-induced gene expression changes were substantially reversed by MLV@MS treatment. We think a small number of differentially expressed genes does not necessarily indicate a weaker therapeutic effect, but may instead suggest higher precision or specificity of the treatment. This means that the intervention regulates only key genes in the inflammatory pathway rather than causing widespread changes across the gene profile. Therefore, we have included a gene heatmap for LPS + MLV@MS vs. LPS (Fig. S20A), which shows that there are significant transcriptional differences between the two groups.Fig. 5Transcriptome sequencing of macrophages after incubation with MLV@MS: (A) Hierarchical clustering heatmap showing differential gene expression profiles among Blank, LPS, and LPS + MLV@MS groups. Red indicates upregulated genes, while blue indicates downregulated genes, clearly illustrating the distinct transcriptional responses to LPS and MLV@MS treatment; (B-D) Volcano plots illustrating differentially expressed genes (DEGs) for: (B) LPS vs. Blank group; (C) LPS + MLV@MS vs. Blank group; (D) LPS + MLV@MS vs. LPS group. Red dots represent significantly upregulated genes, green dots represent significantly downregulated genes (adjusted p < 0.05, |log_2_FoldChange| > 1), and gray dots indicate genes without significant differences; (E, G) Gene Ontology (GO) enrichment analysis for: (E) significantly upregulated genes in the LPS vs. Blank group; (G) significantly downregulated genes in the LPS + MLV@MS vs. LPS group. Categories include biological processes (BP, blue bars), cellular components (CC, red bars), and molecular functions (MF, green bars). MLV@MS markedly attenuated inflammation-related gene expression induced by LPS; (F, H) KEGG pathway enrichment analysis for: (F) significantly upregulated genes in the LPS vs. Blank group; (H) significantly downregulated genes in the LPS + MLV@MS vs. LPS group. Dot sizes correspond to the number of DEGs involved, and color represents statistical significance (p-value). Red boxes highlight key inflammatory pathways, including TNF, IL-17, HIF-1 signaling, and cytokine–cytokine receptor interaction pathways (n = 3). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 5

GO enrichment (Fig. 5E–G) identified that downregulated genes in the LPS + MLV@MS vs. LPS comparison were enriched in “response to stimulus”, “immune system process”, and “cellular anatomical entity”, suggesting suppression of immune activation and inflammatory cascades. Intersection analysis (Fig. S20B–C) further showed that many LPS-upregulated genes were subsequently suppressed by MLV@MS. KEGG analysis (Fig. 5F–H) indicated that MLV@MS inhibited key inflammatory pathways, including TNF-α, IL-17, HIF-1, and cytokine–cytokine receptor interaction. These findings demonstrate that MLV@MS exerts robust anti-inflammatory activity by suppressing cytokine/chemokine expression and key signaling pathways.

P65 is a core subunit of the NF-κB transcription factor complex. Its phosphorylation (p-P65) is a pivotal step in NF-κB activation and a key driver of pro-inflammatory gene expression. Using Western Blot analysis, the p-P65 suppression after MLV@MS treatment was confirmed, which is consistent with the transcriptomic results and indicating inhibition of NF-κB signaling (Fig. S21).

In vivo evaluation of intestinal barrier transport efficiency

2.7

In the in vivo experiments, we first evaluated the ability of orally administered microspheres to enhance the intestinal barrier delivery of MLV. The intestinal delivery capability of calcium alginate microspheres has been extensively demonstrated in previous studies. Here, we conducted a simple evaluation using visible light imaging technology. CV and MLV were labeled with FITC and formulated as CV@MS and MLV@MS, respectively. Following oral administration to mice, the entire gastrointestinal tract, including the stomach, small intestine, and large intestine, was dissected 3 h post-administration and imaged under fluorescence (Fig. 6A). As shown in Fig. 6B and E, MLV@MS exhibited markedly stronger intestinal fluorescence compared to CV@MS, with the signal primarily localized in the small intestine.Fig. 6In vivo oral transintestinal transport evaluation: (A) Schematic diagram of the visible light imaging of in vivo localization after oral administration of MLV@MS in mice (n = 3); (B) Intestinal fluorescence distribution after oral administration of MLV@MS in visible-light imaging analysis; (C) Schematic diagram of in vivo pharmacokinetic evaluation after oral administration of MLV@MS in mice (n = 5); (D) Fluorescence distribution in major organs in visible-light imaging analysis; (E) Semi-quantitative analysis of total intestinal fluorescence intensity; (F) Detection of blood drug concentration in rats after oral administration of MLV@MS in the in vivo pharmacokinetic evaluation; (G) H&E staining of intestinal sections, scale bar = 500 μm; (H) P-gp immunohistochemical staining of intestinal sections, scale bar = 500 μm. (**P < *0.05, ***P < *0.01).Fig. 6

We also compared fluorescence intensity in major organs (heart, liver, spleen, lungs, kidneys), and observed only slight fluorescence in the liver for both microsphere groups, mirroring the trend seen in the intestine (Fig. 6D and Fig. S22). Notably, faint fluorescence from MLV@MS was detected in the knee joints’ anatomical region. These results suggest that calcium alginate microspheres facilitate effective intestinal localization and rapid disintegration in the intestine. While both CV@MS and MLV@MS showed some residual presence in the stomach, the tunable stiffness/charge properties of MLV resulted in significantly stronger intestinal penetration and prolonged intestinal retention. The increased fluorescence in the liver and knee joints for MLV@MS compared to CV@MS indicates superior intestinal penetration and systemic delivery capacity.

We observed fluorescence enrichment of MLV@MS in the knee joint region; however, this alone is not definitive evidence of HAP-1-mediated active targeting. Given the potential for EPR-like passive accumulation at inflamed joints, the current enrichment likely reflects a combined effect of HAP-1 targeting and inflammation-associated EPR. We will further elucidate the role of HAP-1 in future studies through more rigorous, well-controlled experiments. In addition to organ biodistribution and joint targeting, the higher plasma concentration of MTX may also contribute to the therapeutic efficacy against RA. This is consistent with our pharmacokinetic results.

To further assess pharmacokinetic properties, we analyzed the blood concentration of MTX, the MLV@MS payload, using HPLC after collecting rat blood samples at various time points post-oral administration (Fig. 6C–S16). MTX, CV@MS, and MLV@MS were each administered at the same MTX dose (10 mg/kg), and blood samples were collected over 24 h for pharmacokinetic analysis. The MTX concentration was quantified by comparing the HPLC elution curve to a standard (Fig. S23). As shown in Fig. 6F and Table S3, oral administration of MLV@MS substantially altered MTX pharmacokinetics, with a C_max_ of ∼120 ng/mL and a t_1/2_ of 4.6 h. By comparison, MTX and CV@MS reached C_max_ values of 73 ng/mL and 80 ng/mL, and t_1/2_ values of 2.9 h and 3.3 h, respectively. Thus, MLV@MS increased C_max_ by 1.6- and 1.5-fold, and t_1/2_ by 1.6- and 1.4-fold, respectively. These significant pharmacokinetic enhancements are primarily attributable to the improved intestinal barrier penetration facilitated by the microsphere delivery system and the nanostructured core of MLV. Enhanced pharmacokinetics contribute to increased MTX bioavailability and reduced systemic side effects, making MLV@MS a promising strategy for oral drug delivery.

Here, we clarify that we used different species for imaging and pharmacokinetic analyses due to methodological constraints. Our in vivo imaging system supports only gas anesthesia and live imaging in mice; therefore, mice were used. Pharmacokinetic profiling requires serial blood sampling from the same animal, which is not feasible in mice without compromising welfare; thus, rats were used.

Intestinal integrity and pathology were examined by H&E staining (Fig. 6G and S24), which revealed no significant structural changes across groups. The P-gp efflux pump is crucial for intestinal trans-epithelial transport [46]. To further investigate, immunohistochemical staining for the P-gp protein was performed on small intestine sections. Immunohistochemical staining of P-gp in small intestine sections (Fig. 6H) showed markedly increased P-gp expression in the MLV@MS group, followed by CV@MS, with MS also higher than the two controls.

Generally, P-gp overexpression is antagonistic to the intestinal transport of small molecules. However, considering our other relevant findings in this study, we propose that the observed P-gp anomaly may be interpreted as follows. First, upregulation of P-gp in intestinal epithelial cells is a well-established adaptive response to increased luminal exposure and heightened endocytic activity, serving to protect the mucosa by limiting intracellular accumulation of xenobiotics. In our system, MLV@MS enhances mucosal interaction and endocytosis; therefore, elevated P-gp likely reflects compensatory barrier reinforcement rather than a direct predictor of reduced tissue exposure. Second, certain components of MLVs may activate nuclear receptors (PXR/CAR/FXR) or NF-κB, thereby transcriptionally inducing ABCB1. Third, trans-epithelial transport of MLVs primarily proceeds via endocytosis–vesicular trafficking, a route that is essentially P-gp-independent for nanoparticles/macromolecules. Taken together, P-gp upregulation can be viewed as a biomarker of epithelial defense, which does not necessarily compromise delivery efficacy and more likely represents an adaptive response to increased apical stimulation and endocytic activity.

In summary, MLV@MS enhances MLV intestinal delivery via tunable charge and stiffness, promoting efficient mucosal penetration and trans-epithelial transport while activating P-gp-mediated efflux. The improved pharmacokinetic parameters further confirm successful intestinal barrier crossing and prolonged systemic presence of the vesicles. These findings suggest that MLV@MS enables efficient intestinal barrier traversal while preserving the multilayer nanostructure. During penetration, the flexible, positively charged outer layer may shed, exposing the neutral inner layer to promote endocytosis and exocytosis, thus facilitating systemic delivery. Additionally, the enhanced permeability and retention (EPR) effect in inflamed joints, combined with the HAP-1 synovial targeting ligand in the inner core, enables the systemically circulating nanostructures to achieve targeted accumulation in RA-affected joints [[47], [48], [49], [50]].

In vivo evaluation of anti-RA effectiveness and safety

2.8

The above in vivo results preliminarily confirm the intestinal delivery and translocation capability of MLV@MS. We then evaluated its anti-RA therapeutic efficacy in the adjuvant-induced arthritis (AIA) rat model (Fig. 7A). Rat body weight was monitored from week 1 to 5 (Fig. 7B). The MLV@MS group exhibited faster weight gain, closely resembling healthy controls, while the Model and MS groups had slower weight gain. This suggests that oral MLV@MS-mediated RA suppression may improve animal activity and feeding behavior, correlating with body weight gain.Fig. 7In vivo anti-RA efficacy evaluation: (A) Schematic diagram of the evaluation of anti-RA efficacy of MLV@MS in the rat AIA model; (B) Body weight changes in rats; (C) Photographs of rat paws at different time points; (D) Clinical score data for RA symptoms; (E) Paw thickness data; (F) H&E staining in rat ankle sections.; (G) Micro-CT 3D reconstruction and ROI selection of the ankle for bone tissue parameter analysis. The yellow arrow indicates a site of significant bone erosion. (n = 5, **P < 0.01, ***P < 0.001). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)Fig. 7

In the adjuvant-induced arthritis (AIA) rat model, we further evaluated the anti-RA efficacy of MLV@MS on joint symptoms [51]. At each time point, photographs of the animal joints were taken (Fig. 7C), and RA disease scores were assessed using established clinical criteria (Fig. 7D) [52]. Results indicated that the overall RA score in the MLV@MS group rapidly decreased after two weeks, reflecting significant improvement in disease condition. The CV@MS group also showed improvement, though to a lesser extent. Paw thickness measurements (Fig. 7E) further confirmed reduced joint swelling in the MLV@MS group, demonstrating its therapeutic efficacy.

Ankle joint samples were collected, sectioned, and stained with H&E to visually assess joint pathology. The Model group displayed pronounced synovial swelling, while the MLV@MS group showed marked reduction in swelling (Fig. 7F). Pathological changes were further analyzed by Micro-CT, with three-dimensional reconstructions shown in Fig. S25. The calcaneus region was selected as the region of interest (ROI) for detailed analysis of trabecular bone microstructure (Fig. 7G–Table S4). The MLV@MS group exhibited bone parameters most similar to the healthy group, indicating protection against bone erosion. Immunohistochemical staining for IL-1β and TNF-α revealed strong positive staining in the Model and MS groups, while the MLV@MS group exhibited significantly reduced levels of both inflammatory markers (Fig. 8A).Fig. 8In vivo safety evaluation: (A) Immunohistochemical staining of IL-1β and TNF-α in rat ankle sections, scale bar = 1 mm; (B) and (C) Liver function assessment based on ALT and AST levels; (D) and (E) Kidney function assessment based on BUN and CREA levels; (F) Histopathological examination of the heart, liver, spleen, lungs, and kidneys, scale bar = 100 μm.Fig. 8

Collectively, these data demonstrate that oral MLV@MS treatment significantly alleviates RA symptoms, with efficacy superior to CV@MS. As the only difference between CV@MS and MLV@MS is the tunable charge/stiffness conferred by the multilayer vesicles, these results highlight the role of enhanced intestinal transport in improving MTX absorption and anti-RA efficacy. This level of improvement is challenging to achieve with conventional nanomedicine approaches, which may explain the lack of approved MTX or DMARD nanomedicines for RA.

Previous studies have reported that artemisinin and its derivatives, including DHA, possess anti-inflammatory and immunomodulatory activities [28]. Theoretically, these could contribute synergistically with MTX. However, under our dosing regimen, systemic DHA exposure was extremely low, as confirmed by plasma measurements. We included a DHA-containing control group to rule out direct DHA effects on therapeutic outcomes; the results indicated no significant efficacy from DHA alone. Thus, the observed effects are primarily attributable to enhanced MTX absorption via the carrier system, and no clear MTX–DHA synergy was detected in our study.

Although DHA may have therapeutic potential for RA, its rapid release and transport limitations in the intestine led to negligible plasma levels in our experiments, likely due to its short half-life and low stability [6]. While rapid pharmacokinetics suit acute conditions like malaria, they are less favorable for chronic RA therapy, limiting the contribution of orally delivered DHA in this context.

To minimize confounding variables, CV@MS was formulated with an equimolar amount of artesunate (with equivalent biological activity). Despite the potential of artemisinin derivatives, their extremely short half-life results in much lower bioavailability compared to traditional DMARDs such as MTX. Thus, the primary therapeutic effect in RA is due to MTX, with MLV@MS serving as an efficient carrier to enhance its oral delivery.

In addition to efficacy, safety was assessed by monitoring liver and kidney function, as well as potential effects on major organs [53]. At week 5, blood samples were analyzed for ALT and AST (liver function), and BUN and CREA (kidney function) (Fig. 8B–E). All values remained within normal healthy ranges for rats, indicating no adverse effects from microspheres or MLV@MS during the 5-week period. H&E staining of heart, liver, spleen, lung, and kidney sections further confirmed the absence of pathological changes (Fig. 8F). Additionally, intestinal sections from MLV@MS-treated animals showed no structural disruption after five weeks of oral administration, supporting the safety of the system. More comprehensive safety evaluations will be performed in future studies.

16S rRNA sequencing analysis of gut microbiota

2.9

It is well established that gut microbiota dysbiosis is closely linked to the onset and progression of RA. However, indiscriminate elimination of bacterial populations, including beneficial probiotics, can adversely impact the intestinal microenvironment. To evaluate whether and how oral administration of MLV@MS microspheres affects gut microbiota composition and abundance, we performed high-throughput 16S rRNA sequencing analysis of mouse intestinal bacterial communities.

As shown in Fig. 9A, Weighted UniFrac-based non-metric multidimensional scaling (NMDS) revealed that the microbial community of the MLV@MS group clustered closer to that of the Healthy group, indicating a restorative effect on gut microbial homeostasis. In contrast, the Model, MS, and CV@MS groups showed greater divergence from the Healthy group, consistent with a disease or ineffective intervention state.Fig. 916S rRNA sequencing of intestinal microbiota: (A) Weighted UniFrac-based NMDS plot showing the overall differences in gut microbiota composition among experimental groups; (B) Phylogenetic cladogram visualizing LEfSe analysis results; (C) Bar plots of gut microbiota composition at the family level; (D) Notched boxplot of ANOSIM analysis displaying within-group and between-group distances; (E) Genus-level heatmap with hierarchical clustering; (F) Bar chart comparing the relative abundance of major phyla; (G) LEfSe analysis identifying discriminative taxa enriched in MLV@MS or Model groups, reflecting the microbial shifts associated with treatment (n = 5).Fig. 9

A phylogenetic cladogram based on Linear Discriminant Analysis Effect Size (LEfSe) (Fig. 9B) highlighted significant differences in microbial taxa between groups. MLV@MS treatment enriched taxa primarily in the Firmicutes and Actinobacteria phyla, including Turicibacteraceae, Turicibacterales, Christensenellaceae, Corynebacteriaceae, and Micrococcaceae groups associated with immune regulation, intestinal barrier repair, and anti-inflammatory effects. In contrast, the Model group was characterized by increased abundance of potentially pathogenic or pro-inflammatory bacteria. Bar plots at the family level (Fig. 9C) showed restoration of beneficial taxa, such as Lactobacillaceae and Lachnospiraceae, and stable presence of Ruminococcaceae in the MLV@MS group, with an overall microbial profile resembling healthy controls.

ANOSIM analysis (Fig. 9D) further confirmed significant divergence between the MLV@MS and Model groups. Hierarchical clustering and genus-level heatmaps (Fig. 9E) showed increased abundance of Faecalibacterium, a key anti-inflammatory genus, in the MLV@MS group, while pathogenic Staphylococcus levels were reduced. At the phylum level (Fig. 9F), MLV@MS increased beneficial Firmicutes and Actinobacteria, while reducing potentially pro-inflammatory Bacteroidetes, indicating modulation of gut microecology and improved metabolic potential. LEfSe analysis (Fig. 9G) identified significant enrichment of anti-inflammatory and metabolically active taxa (Faecalibacterium, Turicibacter, Actinomyces) in the MLV@MS group, whereas the Model group was dominated by inflammation-associated taxa (Deferribacteraceae, Mucispirillum).

Overall, 16S rRNA sequencing demonstrated that oral administration of MLV@MS effectively modulates gut microbiota structure, increasing beneficial bacteria while reducing pathogenic and pro-inflammatory taxa. The functional potential of the gut microbiota shifted toward anti-inflammatory and metabolically favorable profiles, suggesting that MLV@MS could serve as a therapeutic strategy through both direct intervention and gut microecological modulation.

We attribute these changes primarily to the prebiotic effect of sodium alginate, a marine polysaccharide component of the microsphere matrix. Sodium alginate is partially metabolized by certain gut bacteria, promoting the growth of beneficial microbes and supporting microbiota homeostasis. Previous studies have shown that oral alginate increases the abundance of beneficial genera such as Bacteroides and Lactobacillus, enhancing gut microbial diversity. Additionally, the MLV@MS system helps mitigate the adverse effects of MTX on gut microbiota, as targeted delivery minimizes gut toxicity and preserves microbial balance compared to conventional oral MTX.

Previous work has demonstrated that artemisinin and its derivatives possess anti-inflammatory activity and can modulate gut microbiota [54]. Under our experimental conditions, systemic DHA exposure was negligible, and DHA-related variables were controlled in the relevant groups; thus, the observed microbiota modulation is unlikely to be attributable to DHA. Nevertheless, local release of DHA may exert pharmacological effects, such as directly modulating intestinal immune cells or the epithelial barrier, which could be beneficial for the treatment of rheumatoid arthritis. Dedicated studies are needed to elucidate these local actions and their specific contributions.

The charge- and stiffness-tunable MLV@MS platform described here represents a versatile and responsive oral delivery system. Its environment-sensitive properties make it suitable not only for MTX, but also for a variety of biological macromolecules, including insulin, PTH, and monoclonal antibodies, by improving mucosal interaction and trans-epithelial absorption and protecting drugs from gastrointestinal degradation. While the platform demonstrates broad potential, further optimization and comprehensive evaluation will be required for specific drugs, given the diversity in physicochemical properties and absorption mechanisms.

Gastrointestinal side effects, such as nausea, vomiting, diarrhea, and mucosal injury, are a major limitation of oral MTX therapy in RA. These adverse events affect patient compliance and limit dose escalation, thereby impeding optimal disease control. The main innovation of this work lies in integrating MLVs with microspheres to actively enhance intestinal delivery efficiency for MTX and other RA therapeutics, while reducing gastrointestinal side effects. The MLVs feature a detachable, positively charged, low-stiffness outer shell surrounding a highly stiff core SUV with a low negative surface charge, all encapsulated within a rapidly gelled calcium alginate hydrogel. This architecture ensures storage stability, resistance to gastric digestion, and targeted release in the intestine. The resulting charge/stiffness-tunable structure is precisely tailored for efficient transintestinal transport of DMARDs, providing a promising solution to longstanding challenges in oral nanomedicine for RA.

In future work, we will more deeply investigate the independent functions of each component in the oral delivery system to optimize material design and prepare for clinical translation. Currently, MLV@MS is a multi-component formulation (DAPC, DOTAP, TAT, HAP-1, alginate). We did not include a comprehensive set of negative controls to deconvolute the contributions of individual components (for example, formulations lacking HAP-1 to test targeting or lacking TAT to assess penetration). Owing to feasibility constraints, these experiments were beyond the scope of the present study. In future studies, we will implement systematic component-omission and substitution experiments to clearly delineate the role of each component.

Conclusion

3

In this study, we achieved the property requirements for nanomedicine in transintestinal transport by constructing a nano-microsphere constructed by charge/stiffness-tunable MLV with microsphere encapsulation for RA treatment. This MLV structure includes a negative-charged and high-stiffness SUV core, as well as a positively charged and flexible phospholipid shell based on a charged-tunable phospholipid (DAPC). This tunability of charge and stiffness originates from the DAPC driven continuous charge reversal and self-shedding of the shell during the penetration process. To prevent premature self-shedding, encapsulation with calcium alginate hydrogel microspheres was employed. This pH-responsive hydrogel efficiently delivers MLV to the intestinal environment. In vitro and in vivo experiments clearly demonstrated that this charge/stiffness-tunable MLV combined with the microsphere significantly enhances the intestinal delivery of the MTX payload after oral administration and achieve high anti-RA efficacy. Ultimately, the construction of this novel oral delivery system provides a groundbreaking approach to the design of oral DMARD nanomedicines for RA treatment, addressing challenges that have long perplexed researchers.

Methods

4

Materials

4.1

Sodium alginate was obtained from Sigma-Aldrich (St. Louis, Missouri). Artesunate, dihydroartemisinin (DHA), methotrexate (MTX) and 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-N-hydroxysuccinimide] (DSPE-PEG_2000_-NHS) were supplied by Aladdin (Shanghai, China). 1,2-Distearoyl-sn-glycero-3-phosphocholine (DSPC) and 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) were sourced from Avanti Polar Lipids (Alabaster, Alabama). The di-artesunate-phosphatidylcholine (DAPC) was gifted by Dr. Du from Shanghai Jiao Tong University (Shanghai, China). Cholesterol was purchased from Sigma-Aldrich (St. Louis, Missouri). The TAT (sequence: YGRKKRRQRRR) and HAP1 (sequence: SFHQFARATLAS) peptides were custom-synthesized by Qiangyao Biotech. Co., Ltd (Wuhan, China). A coaxial injection needle (30 + 18G) was custom-made by Ispin Tech. Co. (Hefei, China). Organic solvents, including chloroform and ethanol of analytical grade, were provided by Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China).

Synthesis of DSPE-PEG-TAT and DSPE-PEG-HAP1

4.2

The synthesis of DSPE-PEG-TAT and DSPE-PEG-HAP1 was carried out through the reaction between DSPE-PEG2000-NHS and the TAT or HAP1 peptides [55]. The detailed procedure is as follows: DSPE-PEG2000-NHS and the DMSO stock solution of TAT or HAP1 were added to phosphate-buffered saline (PBS, pH 7.4) at a molar ratio of 10:1 (NHS:peptide). The mixture was stirred at room temperature for 4 h. After the reaction was completed, the resulting DSPE-PEG-TAT and DSPE-PEG-HAP1 products were purified by ultrafiltration using a dialysis membrane with a molecular weight cutoff (MWCO) of 3.5 kDa to remove unreacted peptides and by-products. This was followed by further purification with distilled water. The final products were lyophilized and stored at −20 °C until further use.

LC-MS/MS analysis of DAPC hydrolysis

4.3

After incubation of DAPC with artificial intestinal fluid (AIF) for specific time intervals, the samples were pretreated and analyzed using an LC-MS/MS system (Waters, Milford, MA) equipped with a 15 cm C18 reverse-phase column. The mobile phase consisted of 20 mM ammonium formate buffer (pH 6) as solvent A and acetonitrile as solvent B. The gradient elution program was as follows: 0–2 min, 80% A/20% B; 2–10 min, a linear gradient to 20% A/80% B; 10–12 min, maintained at 20% A/80% B; 12–15 min, returned to 80% A/20% B. The column temperature was maintained at 40 °C, with a flow rate of 0.5 mL/min and an injection volume of 5 μL. Mass spectrometry was performed in positive ion mode using Fourier Transform Mass Spectrometry (FTMS) with Electrospray Ionization (ESI). Full Scan Mode was used to collect signals across the m/z range of 100–1250. Extracted Ion Chromatograms (EICs) were applied to specifically monitor DHA (m/z 267.1578–267.1604), artesunate (m/z 383.1692–383.1730), and DAPC (m/z 990.4408–990.4508). Signal intensities and retention times for DAPC, DHA, and artesunate were recorded at different time intervals (0, 3, 6, 12, and 24 h) to study the degradation process and the formation of products.

Preparation of multilamellar vesicles (MLVs)

4.4

The MLVs were prepared using a double emulsion (W/O/W) method combined with the electrostatically induced self-assembly of DAPC and DOTAP phospholipids, which enabled methotrexate (MTX) loading and the construction of a “onion-like” structure exhibiting stiffness and charge heterogeneity [31].

A sodium bicarbonate solution (10 mM) was used as the inner aqueous phase to maximize the solubilization of MTX. MTX was dissolved at a concentration of 2 mg/mL in this solution. The inner aqueous phase (MTX dissolved in sodium bicarbonate solution) was emulsified into an oil phase containing phospholipids (DSPC and cholesterol at a molar ratio of 3:1, total lipid concentration of 10 mg/mL) dissolved in chloroform to form a water-in-oil (W/O) emulsion (V/V, 1/3). The W/O emulsion was then emulsified into the outer aqueous phase (10 mM PBS buffer, pH 7.4) with DSPE-PEG-HAP1 (molar ratio 1% of DSPC) to form a double emulsion (W/O/W, V/V, 1/3/9). During this process, extensive film homogenization at 50 °C by a liposome extruder (Avanti 610,000, Alabaster, Alabama) was employed to construct small unilamellar vesicles (SUVs). As the pH of the outer aqueous phase was adjusted to neutral (pH 7.4), MTX precipitated within the vesicles, which facilitated the formation of a rigid solid-phase core at the center of the SUVs. After that, DOTAP (100 μg), DAPC (500 μg), and DSPE-PEG-TAT (500 μg) were added to the 5 mL of SUV suspension (∼ lipid concentration of 3.3 mg/mL). After vortexing for 30 min at room temperature). The resulting mixture was subjected to ultrafiltration using a dialysis membrane tube (MWCO 3.5 kDa) to remove unreacted materials and organic solvents to obtain MLVs.

We used a similar preparation method for H-MLV and L-MLV, with the only difference being the substitution of specific phospholipids. For H-MLV, DAPC was replaced with DSPC; for L-MLV, DSPC was replaced with DAPC and cholesterol component was removed as well.

Cryo-EM is used to visualize the multilayered structure of MLVs. Dilute the MLV suspension to 0.2 mg/mL. Cryo-EM samples were prepared by ThermoFisher Vitrobot Mark IV (Hillsboro, OR) and observed by FEI Tecnai G2 Instrument (200 kV, Hillsboro, OR). In addition, the size distribution and zeta potential of MLVs were detected by Malvern NanoZS90 Instrument (Worcestershire, UK).

Preparation of nano-microsphere with calcium alginate microsphere encapsulation

4.5

Dissolve sodium alginate in deionized water to prepare a 2% (w/v) alginate solution [25]. Stir the solution at room temperature for at least 1 h until the alginate is completely dissolved. Filter the alginate solution through a 0.45 μm filter to ensure uniformity and remove any debris. Add the prepared MLVs (at a concentration of 1 mg/mL of total lipids) to the alginate solution. Gently mix the suspension using a magnetic stirrer to ensure homogeneous dispersion of MLVs in the alginate matrix. Load the MLV-alginate suspension into a microfluidic device equipped with a 30 + 18G needle. Load the MLV-alginate suspension into a syringe or microfluidic droplet generator as the flow phase. Use nitrogen gas as the cutting phase to shear the alginate suspension into uniform droplets. Collect the alginate droplets into a 100 mM CaCl_2_ solution under gentle stirring to induce gelation. Droplets instantly form hydrogel microspheres upon contact with the CaCl_2_ solution to obtain final nano-microsphere (MLV@MS). CV@MS was prepared using a similar method, but DAPC was replaced with DSPC and equimolar amount of artesunate with equivalent activity.

The morphology of microspheres was observed by Zeiss microscope (Oberkochen, Germany) and FEI Sirion 200 scanning electron microscopy (SEM, Hillsboro, OR). To evaluate the release rate of MLVs from nano-microsphere, we indirectly characterized it by measuring the amount of MTX released. Briefly, MLV@MS was incubated with AGF, AIF, and ACF (concentration: 10 mg/mL). After incubation for different durations, the samples were centrifuged at 2000 rpm for 3 min at low speed. Once the residual microspheres had settled, the supernatant was collected and analyzed using HPLC to calculate the total release percentage.

We also calculated the encapsulation efficiency and drug loading of MTX in MLV@MS by directly quantifying MTX content. Add lyophilized MLV@MS (10 mg) to 50 mM Tris-HCl buffer (pH 8.0) and incubate at room temperature with gentle shaking for 10 min. Subsequently, add Triton X-100 to a final concentration of 1% (v/v) and incubate for 5 min to lyse the lipid components. Then add an equal volume of acetonitrile/methanol (1:1, v/v) to achieve an organic content of approximately 50% (v/v), sonicate in a water bath for 10 min, and vortex for 60 s to fully extract MTX. Centrifuge at 10,000 g for 15 min at 4 °C, collect the clear supernatant, and filter through a 0.22 μm filter prior to HPLC analysis (Agilent Technologies, Inc., Santa Clara, CA). MTX content is determined using a matrix-matched calibration curve, from which DL and EE are calculated as: Drug Loading (DL) = (mass of MTX in MLV@MS)/(total mass of MTX) × 100%; Encapsulation Efficiency (EE %) = (mass of MTX in MLV@MS)/(total input mass of MTX) × 100%.

In vitro live/dead staining and cytotoxicity assay

4.6

Human colorectal adenocarcinoma cell line (Caco-2) was cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin. The cells were maintained at 37 °C in a humidified atmosphere with 5% CO_2_. Blank MS, CV@MS, and MLV@MS formulations were prepared in complete culture medium at a final concentration of 10 μg/mL. The culture medium in each well was replaced with the respective MS-containing medium, and the cells were incubated for 24 and 48 h at 37 °C.

To evaluate the cytotoxicity of the formulations, live/dead staining was performed using Calcein-AM/PI kit (Beyotime Biotechnology, Shanghai, China). After treatment, the cells were washed twice with PBS to remove excess formulations. A staining solution containing 2 μM Calcein-AM and 5 μg/mL PI was prepared in PBS. A total of 200 μL staining solution was added to each well, and the plate was incubated at 37 °C for 15 min. Following incubation, the cells were gently washed once with PBS and observed under a fluorescence microscope. Green fluorescence (Calcein-AM) indicated live cells, while red fluorescence (PI) indicated dead cells.

To evaluate the cytotoxicity and proliferation behavior of Caco-2 and mouse fibroblast (L929) cells exposed to microsphere formulations, the CCK-8 assay was performed. Caco-2 and L929 cells were cultured in 96-well plates at a density of 5 × 10^3^ cells/well in 100 μL of complete culture medium. After 24 h, the cells were treated with blank MS, CV@MS, or MLV@MS at a final concentration of 10 μg/mL. Untreated cells served as the control group. At incubation of additional 1, 3 and 5 days, 10 μL of CCK-8 reagent was added to each well and incubated for 1 h at 37 °C. The optical density (OD) at 450 nm was measured using a microplate reader (Thermo Fisher Scientific, Waltham, Massachusetts). The OD values reflected the relative number of viable cells in each well, providing an indirect measure of cytotoxicity and cell proliferation.

In vitro flow cytometry analysis

4.7

To evaluate the uptake efficiency of different vesicles by Caco-2 cells, flow cytometry was performed. CV (SUV) and MLV were pre-labeled with the fluorescent dye 3,3′-Dioctadecyloxacarbocyanine perchlorate (DiO). As a control, an equivalent amount of DiO was dissolved in PBS buffer to prepare a DiO suspension, and its supernatant was used as the DiO Control.

DiO-labeled CV and MLV were prepared by incorporating DiO into the lipid bilayer during vesicle fabrication. The excess DiO was removed by ultracentrifugation and washing with PBS to ensure only vesicle-bound DiO remained. Caco-2 cells were seeded into 6-well plates at a density of 2 × 10^5^ cells/well. The cells were incubated with DiO-labeled CV, MLV, or the DiO Control at equivalent concentrations of DiO for 4 h at 37 °C. After incubation, cells were washed three times with cold PBS to remove unbound vesicles or dye. The cells were detached using 0.25% trypsin-EDTA and resuspended in cold PBS containing 2% FBS. The cell suspension was centrifuged at 300 g for 5 min, and the pellet was resuspended in 500 μL of PBS for flow cytometry analysis. The fluorescence intensity of DiO was measured using a flow cytometer (BD, San Jose, California). The mean fluorescence intensity (MFI) of DiO-positive cells was used to quantify vesicle uptake efficiency. In addition, various endocytosis inhibitors were included as comparators. Cellular uptake was quantified by flow cytometry based on DiO fluorescence.

In vitro cellular uptake analysis under CLSM

4.8

To visually compare the cellular uptake of CV@MS and MLV@MS, CLSM was utilized. DiO-labeled CV@MS and MLV@MS were prepared by encapsulating DiO-labeled vesicles into microspheres. Caco-2 cells were seeded into glass-bottom dishes at a density of 1 × 10^5^ cells/dish and cultured for 24 h. Cells were incubated with CV@MS, MLV@MS, or the DiO Control (supernatant of the DiO suspension) for 4 h at 37 °C. After incubation, cells were washed three times with PBS to remove unbound particles or dye. Cells were fixed with 4% paraformaldehyde for 15 min at room temperature and washed with PBS. Nuclei were counterstained with DAPI (1 μg/mL) for 5 min to visualize cell nuclei. The cells were imaged using a confocal laser scanning microscope (Zeiss, Oberkochen, Germany). CLSM images were processed using ImageJ software to visualize fluorescence localization and intensity.

Ex vitro evaluation of intestinal barrier transport efficiency

4.9

To evaluate the intestinal barrier-crossing ability of MLV@MS, an ex vivo small intestine model was employed. Male Sprague-Dawley rats (200∼250 g) were sacrificed according to institutional guidelines for animal care and use. A segment of the small intestine (∼3 cm in length) was carefully excised and washed with ice-cold PBS to remove residual contents. Both ends of the intestinal segment were ligated using sterile sutures, creating a sealed lumen. A mixed suspension of artificial intestinal fluid (AIF) and MLV@MS (concentration of 10 mg/mL) was prepared and used to fill the lumen of the ligated intestinal segment. For comparison, DiO-labeled control formulations, including DiO supernate and CV@MS with the same DiO concerntration (∼1 μg/mL), were also prepared and introduced into separate intestinal segments. The prepared intestinal segments were immersed in 5 mL of external culture medium in a sterile container. The system was incubated at 37 °C for 1 h under gentle shaking to simulate intestinal conditions.

After incubation, the intestinal segments were harvested, rinsed with cold PBS, and embedded in optimal cutting temperature compound. Frozen sections were prepared using a cryotome and mounted onto glass slides. The frozen sections were fixed briefly with 4% paraformaldehyde for 10 min at room temperature. The sections were washed three times with PBS and stained with DAPI (1 μg/mL) for 5 min. The stained sections were observed using a CLSM (Zeiss, Oberkochen, Germany).

The external culture medium surrounding each intestinal segment after the 1-h incubation was collected and filtered to remove debris. This medium was then applied to Caco-2 cells (seeded in glass-bottom dishes at 1 × 10^5^ cells/dish) for 1 h at 37 °C. After incubation, the Caco-2 cells were washed with PBS, fixed with 4% paraformaldehyde, and stained with DAPI for CLSM observation.

To assess vesicle-mucus interactions and penetration dynamics, porcine gastric mucin was used to prepare a simulated mucus layer evenly coated on glass-bottom plates and equilibrated at 37 °C. DiD-labeled multilamellar vesicles (DiD-MLV) and conventional vesicles (DiD-CV) were then added (lipid concentration 1.5 μg/mL; 100 μL per well). After removal of free dye, samples were co-incubated at 37 °C. z-stack images were acquired by confocal microscopy (z range 0∼2 μm; time points 0, 7, and 14 min), followed by 3D reconstruction and generation of orthogonal views to visualize penetration. The intensity inflection point was defined as z = 0, and DiD fluorescence depth was calculated at each time point to obtain penetration depth-time profiles. For each group, at least 3 fields of view were analyzed.

In vitro scratch assay

4.10

To assess the effect of MLV@MS on the migration of RA synovial fibroblasts (RA-FLS), a scratch assay was performed using the MH7A cell line. MH7A cells were seeded into 6-well plates at a density of 5 × 10^5^ cells/well. A sterile 200 μL pipette tip was used to create a straight scratch (wound) in the cell monolayer. The detached cells were gently washed away with PBS. The cells were treated with: Control group: No treatment (only culture medium). MS group: Empty microspheres (100 μg/mL). CV@MS group: Microspheres loaded with MTX in CV (100 μg/mL, with a calculated MTX concentration of 200 ng/mL). MLV@MS group: Microspheres loaded with MTX in MLV (100 μg/mL, with a calculated MTX concentration of 200 ng/mL). The final volume in each well was 2 mL of culture medium. All treatments were incubated for 24, 48, and 72 h at 37 °C. At each time point, the scratch wounds were imaged using a microscope (Zeiss, Oberkochen, Germany). The wound area was measured using ImageJ software to quantify the remaining scratch area.

In vitro evaluation of macrophage inflammatory polarization

4.11

To assess the effect of MLV@MS on macrophage inflammatory polarization, LPS-induced RAW264.7 cells were treated with different formulations, and polarization markers were evaluated using immunofluorescence staining, qPCR, and ELISA. RAW264.7 macrophages were seeded into 24-well plates with glass coverslips at a density of 1 × 10^5^ cells/well in complete DMEM. Cells were stimulated with LPS (100 ng/mL) and incubated for 6 h to induce M1 polarization. Cells were treated with: Control group: No treatment. MS group: Empty microspheres (100 μg/mL). CV@MS group: Microspheres loaded with MTX in CV (MTX concentration: 200 ng/mL). MLV@MS group: Microspheres loaded with MTX in MLV (MTX concentration: 200 ng/mL). Treatments were performed for 24 h at 37 °C. After incubation, cells were washed three times with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature. Cells were permeabilized with 0.1% Triton X-100 in PBS for 10 min and blocked with 3% BSA for 30 min to prevent nonspecific binding. Cells were incubated with primary antibodies for iNOS-FITC (M1 marker) and CD206-PE (M2 marker) at 4 °C overnight. After washing three times with PBS, cells were incubated with appropriate fluorescent secondary antibodies for 1 h at room temperature. Cells were counterstained with DAPI for 5 min. Fluorescence images were captured using a CLSM (Zeiss, Oberkochen, Germany).

To assess the effect of MLV@MS on macrophage inflammatory polarization, the gene and protein expression levels of inflammatory factors (IL-1β and TNF-α) were measured using qPCR and ELISA, respectively. RAW264.7 macrophages were seeded into 6-well plates at a density of 2 × 10^5^ cells/well and cultured in complete DMEM. Cells were stimulated with LPS (100 ng/mL) for 6 h to induce M1 polarization. After stimulation, the cells were treated with the following groups for 24 h. After incubation, total RNA was extracted from the cells using TRIzol reagent or a commercial RNA extraction kit, according to the manufacturer's instructions. RNA concentrations were quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, Massachusetts), and RNA quality was assessed. RNA (1 μg) was reverse-transcribed into complementary DNA (cDNA) using a reverse transcription kit. qPCR was conducted according the well-established protocol. The relative expression levels of IL-1β and TNF-α were normalized to the housekeeping gene using the 2^−ΔΔCt^ method.

For ELISA assay, RAW264.7 macrophages were seeded into 12-well plates at a density of 1 × 10^5^ cells/well. After 24 h of treatment, the culture supernatant from each well was collected and centrifuged at 300 g for 5 min to remove cell debris. The supernatants were stored at −80 °C until analysis. Cytokine levels of IL-1β and TNF-α in the supernatants were measured using commercial ELISA kits (PeproTech, Suzhou, China), following the manufacturer instruction.

In vivo visible light imaging analysis

4.12

All the animal experiments were approved by the Animal Welfare and Ethics Committee of PNK (Shanghai) Co., with the ethical approval number SYXK (Hu) 2023-0038. To investigate the intestinal barrier transport efficiency of MLV@MS, we performed in vivo experiments using a fluorescence imaging-based approach. FITC was used to label MLV, and MLV@MS. Male BALB/c mice (18∼22 g) were used in this study. All animal experiments were approved by the institutional animal care and use committee. The mice were fasted for 4 h before the experiment but allowed free access to water. The animals were randomly divided into two groups for analysis at different time points (1 and 3 h post-administration, n = 3). Each mouse was orally administered FITC-labeled MLV@MS suspension (containing 10 mg of MS) via gavage. The mice were sacrificed by cervical dislocation under isoflurane anesthesia. The entire digestive tract, including the stomach, small intestine, cecum, and large intestine, was carefully dissected and washed with PBS to remove residual contents. Each segment was placed on a black background to minimize autofluorescence. The tract segments were imaged using a IVIS fluorescence imaging system (Perkin Elmer, Waltham, Massachusetts).

In vivo pharmacokinetic analysis

4.13

To evaluate the impact of MLV@MS on intestinal transmembrane penetration efficiency and the pharmacokinetics of its payload, methotrexate (MTX), blood concentration measurements were conducted using high-performance liquid chromatography (HPLC). Briefly, adult male rats (200∼250 g) were divided into three groups and orally administered of Free MTX, CV@MS and MLV@MS with the equal MTX dose of 10 mg/kg (n = 5). Blood samples (approximately 200 μL) were collected from the orbital vein at predetermined time intervals. Blood samples were stored in heparinized tubes to prevent coagulation. Blood samples were centrifuged at 4000 rpm for 10 min at 4 °C to separate plasma. To remove plasma proteins, 100 μL of plasma was mixed with 300 μL of acetonitrile. The mixture was vortexed for 2 min and centrifuged at 10,000 rpm for 10 min. The supernatant was collected and filtered through a 0.22 μm membrane filter for HPLC analysis. HPLC with a C18 reverse-phase column (4.6 × 150 mm, 5 μm) (Agilent, Santa Clara, California) was used for detection. Mobile phase was a mixture of phosphate buffer (pH 6.8) and acetonitrile in a 70:30 ratio. Flow rate was 1.0 mL/min. The area under the MTX elution peak was integrated and compared to a curve prepared using MTX standard sample. The plasma concentration-time data were analyzed using a one-compartment pharmacokinetic model.

The selection of DHA for DAPC synthesis was primarily based on its well-established safety profile and anti-inflammatory properties. Although DHA did not demonstrate a significant synergistic therapeutic effect in the present system, its known anti-inflammatory and immunomodulatory activities suggest potential for the future development of multi-targeted therapeutic strategies. Thus, DHA was chosen as one of the candidate molecules for functionalization. We believe that the anti-inflammatory properties of DAPC could be beneficial for RA therapy, but broader applications in disease treatment will require further investigation and validation.

Establishment of rat adjuvant-induced arthritis (AIA) model

4.14

The adjuvant-induced arthritis (AIA) model was established in male Sprague-Dawley rats, which is widely used to study the pathophysiology of rheumatoid arthritis (RA) and evaluate potential therapeutic agents [51]. Male rats (180∼220 g) were used for the study. The rats were housed under standard conditions with a 12-h light/dark cycle and free access to food and water. Complete Freund's Adjuvant (CFA) was prepared by suspending heat-killed Mycobacterium tuberculosis (10 mg/mL) in paraffin oil. The CFA solution was vortexed thoroughly to ensure homogeneity before use. On Day 0, each rat was anesthetized with isoflurane to minimize discomfort. A single dose of 0.1 mL CFA was injected into the subcutaneous tissue of the left hind paw. Rats were monitored daily for signs of arthritis, including paw swelling, erythema (redness), and joint stiffness. After 7 days post-injection, AIA model was established, and increase in paw thickness was used as an indicator of inflammation severity.

AIA rats were randomly divided into four groups (n = 5) and were orally administered PBS, MS, CV@MS, and MLV@MS, respectively. The oral gavage was performed once every three days using the same dose of MTX (20 mg/kg), and the mass of blank MS microspheres was consistent with that of CV@MS and MLV@MS microspheres. In addition, a group of healthy rats was set as a control for comparison. Body weight and paw thickness were measured once a week. After six weeks of oral gavage, the rats were sacrificed for sample collection.

The clinical severity of rheumatoid arthritis in rats was evaluated using a 5-point scoring system based on the degree of paw swelling and joint deformity. The scoring criteria are as follows [10]: points: No swelling or redness; 1 point [2]: Slight swelling and/or redness in a single joint [32]; points: Moderate swelling and redness involving multiple joints [43]; points: Severe swelling with limited joint mobility [54]; points: Severe swelling and deformity with significant joint stiffness or immobility; (6) 5 points: Maximum severity, including extreme swelling, deformity, and complete loss of joint function.

Histopathological analysis

4.15

Histopathological examination was performed to evaluate the joint tissue damage and inflammation in the AIA rat model. Intestinal segments were fixed immediately in 4% paraformaldehyde. After fixation, rinse the tissue with PBS to remove residual fixative. Dehydrate the tissue using a graded ethanol series (70%, 80%, 90%, 95%, and 100%, 1∼2 h per step). Clear the tissue in xylene. Embed the tissue in paraffin wax and allow it to solidify.

For paw samples, carefully remove the hind paw by cutting above the ankle joint using a scalpel or fine scissors. Ensure the entire ankle joint, including surrounding soft tissues (synovium, cartilage, and bone), is intact. Rinse the paw with cold phosphate-buffered saline (PBS) to remove blood and debris. Immerse the tissue in 4% paraformaldehyde. Then immerse the fixed paw in the decalcification solution at room temperature for 4 weeks, depending on the size and density of the bone. Replace the decalcification solution every 2∼3 days to ensure continuous decalcification. Dehydrate the tissue using a graded ethanol series (70%, 80%, 90%, 95%, and 100%, 1∼2 h per step). Clear the tissue in xylene. Embed the tissue in paraffin wax and allow it to solidify.

Paraffin block and sectioning were conducted subsequently. Then, hematoxylin and eosin (H&E) staining and immunohistochemistry were used. In addition to the ankle joint analysis, the heart, liver, spleen, lungs, kidneys, and other organs undergo similar paraffin embedding, sectioning, and H&E staining for histopathological examination. Meanwhile, collect blood samples and centrifuge the blood at 3000 rpm for 10∼15 min to separate plasma. Send the plasma or serum to a clinical laboratory for biochemical analysis of liver (ALT, AST) and kidney function (BUN, creatinine).

Micro-CT analysis of ankle samples

4.16

Ankle specimens from each group of rats were carefully dissected to remove surrounding soft tissues and fixed in 4% paraformaldehyde at 4 °C for 48 h. The fixed samples were then scanned using a Micro-CT scanner (Skyscan 1176, Bruker, Antwerp, Belgium), with the following parameters: voxel size of 9 μm, voltage of 50 kV, and current of 200 μA. The region of interest (ROI) covering the ankle joint was selected for quantitative analysis using the manufacturer software [56]. Key parameters, including bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation (Tb.Sp), were calculated to evaluate the microarchitecture of trabecular bone.

16S rRNA sequencing of intestinal microbiota

4.17

Using 16S rRNA sequencing, we analyzed the intestinal microbiota of mice to assess changes in bacterial composition and abundance following treatment with PBS, MS, CV@MS, MLV@MS. Healthy mice were used as control. On the 7th day, fecal samples of mice were collected. Briefly, total genomic DNA was isolated by Stool Genomic DNA Extraction Kit (Solarbio, Shanghai, China) and amplified by a GeneAmp® 9700 thermal cycler (Applied Biosystems, Foster City, CA). Sequencing libraries were constructed with the Illumina TruSeq DNA kit and sequenced on the Illumina MiSeq platform according to the manufacturer's instructions. Raw sequencing data were processed using QIIME2 software. Weighted UniFrac NMDS, ANOSIM, LEfSe, and hierarchical clustering analyses were conducted to assess differences in microbial community structure among groups.

Statistical analysis

4.18

All data are expressed as mean ± standard deviation (SD). For comparisons between two groups, an independent samples t-test was used. P < 0.05 was considered statistically significant.

Declaration of AI assistance

The author team independently developed this manuscript and employed ChatGPT exclusively for linguistic improvement after draft completion. Following this assistance, the authors reviewed and edited all content, accepting full responsibility for the final publication.

CRediT authorship contribution statement

Yingchun Zhu: Funding acquisition, Investigation, Methodology, Writing – original draft. Lei Wang: Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Writing – original draft. Yin Zhang: Data curation, Formal analysis, Investigation. Guanrong Li: Data curation, Investigation. Zheyuan Shi: Investigation. Dianqing Wang: Methodology. Qiang Wang: Methodology. Xin Yang: Software. Liheng Wang: Investigation, Supervision, Writing – review & editing. Jun Shen: Project administration, Supervision, Writing – review & editing. Xing Yang: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing – review & editing.

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.