Investigation of PSBMA‐PU Ternary Copolymer‐Based Semipermeable Membrane for Macro‐Encapsulation of Cells

ABSTRACT

Cell macro‐encapsulation necessitates semi‐permeable membrane materials with strong resistance to fibrotic encapsulation in vivo. In this study, segmented polyzwitterionic‐polyurethane block copolymers (SPU) were synthesized to meet this requirement. Fourier‐transform infrared and X‐ray photoelectron spectroscopy confirmed the molecular structure of the successfully synthesized SPU. When incorporated into commercial polyurethane, SPU significantly reduced the hydrophilicity of electrospun fiber membranes and exhibited minimal protein adsorption. The resulting membranes demonstrated favorable mechanical properties, well‐defined porous architecture, and efficient transmembrane mass transfer. Encapsulation experiments using BMSCs confirmed that the membrane effectively retained cells while supporting normal cellular growth, allowing secreted factors to diffuse through the membrane and induce M1‐to‐M2 macrophage polarization. In vivo subcutaneous implantation in rats showed that membranes containing the zwitterionic block copolymer elicited significantly thinner fibrous capsules than both unmodified polyurethane and PTFE controls at 14 and 28 days, with no signs of acute or chronic inflammation.

In this study, the synthesis and electrospinning process of polyzwitterionic‐polyurethane block copolymer (SPU) are described, and its physical and chemical properties are characterized. Through cell experiments and subcutaneous implantation experiments in rats, the anti‐protein adsorption, anti‐inflammatory, and inhibition of fiber encapsulation effects of the materials are verified, which provided support for the development of cell macro encapsulation carriers.

1. Introduction

Cell therapy holds significant clinical potential for treating degenerative disorders, autoimmune diseases, and certain cancers [1]. The therapeutic effects of cell‐based therapy are largely mediated by paracrine secretion of bioactive molecules thereby facilitating endogenous tissue repair [2]. However, direct in vivo application of cells faces risks include immune rejection, teratoma, or tumor formation [3]. Cell encapsulation is a promising method to decrease the risks [4].

Micro‐encapsulation is to embed cells in a thin spherical semi‐permeable polymer layer, forming particle diameters of 100–600 µm. Sodium alginate is the mainstream material for mesenchymal stem cells (MSC) encapsulation, which can reduce the mechanical damage of MSCs by injection [5]. By adjusting the microcapsule size and the thickness of the capsule wall, the mass exchange rate of nutrients and oxygen can be optimized to ensure the vitality of MSC, and the vitality of the embedded cells can be preserved. The hydrophilic polymer with excellent biocompatibility including sodium alginate, cellulose, collagen, and polyethylene glycol (PEG), is generally used for preparing microcapsules [3, 6, 7]. However, the polymer membrane layer is prone to uneven layer thickness and pore size. Inconsistent performance across batches regarding mass transfer efficiency is hard to avoid [8].

Macro‐encapsulation uses a semi‐permeable membrane capsulating large number of cells. That avoids the shortcomings of micro‐encapsulation since the membrane where cells are strictly confined can be uniformly fabricated. Macro‐encapsulation creates a localized hypoxic microenvironment, which not only maintains or enhances cell secretion but also promotes peri‐device vascularization, enabling prolonged capsulated cell activity.

Nevertheless, key challenges of fibrotic encapsulation still need to be overcome [9, 10, 11]. Polyurethane (PU) is a favorable material for encapsulation membranes [2] due to its excellent mechanical strength [12], physical stability, and biocompatibility. However, inherent hydrophobicity of PU impedes mass exchange and predisposes to foreign body reactions and fibrosis [13, 14]. As such disadvantages aforementioned, it is necessary to develop hydrophilic and low‐irritating polyurethane to improve the performance and make up for the shortcomings.

Zwitterionic polymers, featuring equimolar cationic and anionic groups, exhibit high hydrophilicity and strong resistance to biofouling. Incorporating zwitterionic structures into polyurethane can enhance hydrophilicity, reduce mass transfer resistance, suppress protein adsorption, and minimize foreign body reactions and fibrosis. Current approaches often rely on surface modification techniques—such as grafting or coating—which are inefficient, may block pores, and compromise membrane integrity [15, 16, 17, 18]. In contrast, integrating hydrophilic zwitterionic segments directly into the polyurethane backbone enables uniform modification without structural disruption, offering a robust route to high‐performance porous membranes.

In this study, segmental zwitterionic polyurethane was synthesized via chain extension. [BMIM]Cl and dimethyl sulfoxide (DMSO) were employed to improve miscibility between amphiphilic block copolymers and commercial PU, enabling fabrication of semi‐permeable membranes with tunable pore size by electrospinning. The resulting membranes were systematically evaluated for morphology [19], porosity [20], hydrophilicity [19], anti‐protein adsorption capacity [21], water flux [22], mechanical properties [23], and in vitro compatibility with cells and tissues [24]. Modified membranes were further fabricated into capsule bags to assess cellular proliferation and functional expression. In vivo fibrotic encapsulation of the membranes was also examined through animal experiments.

2. Results

2.1. Chemical Structures of the Copolymer

The synthetic route of PSBMA‐OH and the block copolymers is illustrated in Figure 1a. pPU prepolymers were coupled with PSBMA‐OH via nucleophilic addition to yield the corresponding segmented block copolymers SPU. Notably, SPU was synthesized without a chain extension step, resulting in a higher molar fraction of PSBMA within the polymer architecture.

FIGURE 1.

(a) Synthesis route of polyurethane block copolymer; (b) IR spectrum of polymer PU and its copolymer; (c) surface element scanning full spectrum; (d) Narrow spectrum of S element; (e) N‐element peak fitting graph; (f) The ¹H NMR spectra of PU and SPU; (g) GPC spectrum of PU.

Fourier transform infrared spectroscopy (FT‐IR) analysis (Figure 1b) revealed that SPU exhibits characteristic absorption bands attributable to polyurethane moieties, including N─H stretching (3320 cm^{− 1}), C═O stretching (1700 cm^{− 1}), and C─N stretching (1530 cm^{− 1}). Additionally, a distinct vibrational mode at 1110 cm^{− 1} corresponds to the asymmetric stretching of ${\mathrm{SO}}_3^{-}$ groups in the zwitterionic sulfobetaine unit. The broad absorption band centered at 3400 cm^{− 1} is assigned to O─H stretching vibrations arising from hydroxyl end groups and bound water associated with PSBMA, with enhanced intensity in SPU due to its higher PSBMA content.

X‐ray photoelectron spectroscopy (XPS) analysis (Figure 1c) confirmed the presence of four elemental components on the SPU surface: C1s (288.11 eV), N1s (399.27 eV), O1s (532.2 eV), and S2p (167.8 eV). High‐resolution S2p spectrum (Figure 1d) displays a doublet signal consistent with the sulfur atom in the sulfonate group of PSBMA. The deconvoluted N1s spectrum (Figure 1e) exhibits two distinct nitrogen species: the peak at 400.78 eV is attributed to quaternary ammonium nitrogen (N⁺) in the sulfobetaine structure, while the signal at 399.44 eV corresponds to the secondary amine nitrogen in the urethane linkage, confirming the coexistence of two chemically distinct nitrogen environments in SPU.

It can be seen from the nuclear magnetic hydrogen spectrum (Figure 1f) that there are more peaks than the PU structure in the chemical shift of 4.04, 3.93, 3.34, and 2.95 ppm, which is the chemical shift of the hydrogen of the introduced PSBMA. PSBMA was introduced into PU.

Because there is no solvent that can dissolve SPU in the commonly used GPC test, only the molecular weight of the PU segment in the block copolymer is evaluated here. It can be seen from Figure 1g that the outflow time and peak time of the two polyurethanes are the same, and the molecular weight distribution is narrow. The molecular weight of the polyurethane is measured to be about 220 000. Therefore, the polyurethane synthesized in this experiment only has different hard segment content. This setting is to adjust the hard segment ratio of the PU segment in the block copolymer and improve the solubility of the polyurethane segment in DMSO and ionic liquid solvents with high polarity.

The diameter, pore size and porosity of the fiber membrane play a crucial role in the exchange of materials. High porosity is more beneficial for material exchange. In order to explore the related properties of the fiber membrane, we analyzed the surface morphology of the fiber membrane. The surface morphology of the electrospun fibrous membrane fabricated from a blend solution of SPU and commercial PU is presented in Figure 2a. The electrospun fiber membranes containing SPU weights of 0%, 20%, 35%, and 50% are coded as EM‐0, EM‐20, EM‐35, and EM‐50. Incorporation of the block copolymer SPU did not markedly alter the fiber surface topography, yet resulted in a reduction in fiber diameter with increasing SPU content. The average fiber diameters of EM‐0, EM‐20, EM‐30, and EM‐50 were founded to be 2.415, 1.459, 1.305, and 0.579 µm, respectively. Consequently, both the average pore size and porosity arising from fiber mat stacking progressively decreased. The coarser fibers in EM‐0 exhibited a more loosely packed architecture with higher porosity (85.4%), whereas the EM‐50 membrane, possessing the smallest fiber diameter, displayed the most compact structure and a reduced porosity of 81.5%, representing a 3.9% decrease relative to EM‐0.

FIGURE 2.

SEM images of electrospinning fiber membranes and Electrospinning membranes with different components (a) SEM images of electrospinning fiber membranes; (b) Fiber diameter. Number of samples (n = 5); (c) Fiber membrane porosity. Number of samples (n = 3); (d) Stress–strain curve of electrospinning fiber membrane. Number of samples (n = 3). The significance analysis used above is an ordinary one‐way analysis of variance. The specific post‐hoc test used above is Dunnett.

The mechanical properties of fibrous membranes are crucial to the stability of cell encapsulation devices. In order to determine whether the strength of the fiber membrane meets the requirements of cell encapsulation, we tested the tensile‐strain curve of the fiber membrane. The influence of SPU incorporation on the mechanical performance of the membranes is illustrated in the tensile stress–strain curves shown in Figure 2d. With increasing SPU content, both the elongation at break and tensile strength declined. At 50 wt.% SPU loading, the tensile strength diminished from 11.93 to 6.19 MPa, while the elongation at break dropped from 370.79% to 185.92%, indicating an approximate 50% reduction in mechanical strength. Despite this decline, the resulting mechanical properties remain superior to those of conventional hydrogel matrices used for cell encapsulation, thus providing adequate structural integrity for macro encapsulation applications. Compared with the existing research, the mechanical strength of the zwitterionic fiber membrane prepared by Zhang et al. decreased to 21.8% after grafting SBMA [25]. The zwitterionic polyurethane prepared in this paper has better mechanical properties.

Figure 3a presents the dynamic water contact angles of the four electrospun fiber membranes. The initial water contact angle of EM‐0, which lacks the hydrophilic soft segment SPU, is approximately 120°, indicating a relatively hydrophobic surface. Upon contact, water droplets gradually infiltrate the membrane over ∼20 s, demonstrating limited wettability. With the incorporation of hydrophilic SPU, the contact angle decreases significantly, accompanied by accelerated water permeation kinetics. At the initial measurement stage, the three SPU‐containing membranes exhibit comparable contact angles; however, the permeation rate increases with higher SPU content. Complete water penetration through EM‐50 occurs 7 s faster than through EM‐20, which is attributed to enhanced surface hydrophilicity and improved capillary action arising from reduced pore dimensions.

FIGURE 3.

(a) Dynamic water contact angle of fiber membrane; (b) Protein adsorption on the surface of fiber membranes. Number of samples (n = 3); (c) Water flux measuring device for microporous membrane; (d) Fiber membrane water flux. Number of samples (n = 3); (e) In vitro degradation study of SPU. The significance analysis used above is an ordinary one‐way analysis of variance. The specific post‐hoc test used above is Dunnett.

Protein adsorption will cause certain inflammatory reactions and fibrosis, which will affect the effectiveness of the packaging device. In order to explore the ability of fiber membrane to resist protein adsorption, we tested the anti‐protein adsorption of fiber membrane. Figure 3b displays protein adsorption behavior on different fiber membrane surfaces. Relative to EM‐0, the membrane containing 20% SPU shows no significant change in protein adsorption capacity. In contrast, membranes with 35% and 50% SPU exhibit markedly reduced protein adsorption, suggesting effective resistance to non‐specific protein adhesion due to increased surface hydrophilicity and improved hydration layer formation. This has similar properties to the zwitterionic modified polyurethane studied by Zhang et al. [25].

Figure 3c illustrates the experimental setup for water flux measurement, where water flux serves as a critical metric for evaluating membrane permeability and mass transfer efficiency under convective flow conditions. As shown in Figure 3d, the EM‐35 membrane achieves the highest water flux at 4361.4 L·m^{− 2}·h^{− 1}, surpassing EM‐0. Although EM‐50 demonstrates the fastest permeation under static wetting conditions, its dynamic water flux (3187.3 L·m^{− 2}·h^{− 1}) is lower than that of EM‐35. This discrepancy is likely due to a denser fibrous morphology resulting in elevated hydraulic resistance during filtration, which counteracts the beneficial effects of hydrophilicity and capillary‐driven transport [26].

The degradation performance of modified polyurethane is also very important for cell encapsulation devices. In the existing research Zhang et al. showed that the polyurethane polymerized with SBMA swelled significantly in polar solvents such as methanol, acetone, and DMF at room temperature for 10 days, while only slight swelling occurred in non‐polar solvents such as water and petroleum ether. After being soaked in phosphate buffered saline solution, incubated at 37°C for 14 days, the mass retention rate was 99.16%, indicating that it had excellent stability, and no obvious degradation or mass loss occurred after short‐term implantation [25]. In the in vitro degradation experiments on SPU, the material was immersed in PBS solution and tested for accelerated degradation at 50°C (Figure 3e). By day 18, it still retained approximately 80% of its mass.

2.2. Experimental Study on Cell Encapsulation

Figure 4a illustrates the cell encapsulation pockets fabricated using a fibrous membrane (B‐EM‐50) and a dense membrane (B‐Dense), respectively, derived from the same synthesized SPU/commercial PU composite. One terminus of each pouch was interfaced with a catheter for intracapsular injection of BMSCs, followed by sealing with medical‐grade adhesive to ensure hermetic encapsulation. Figure 4b presents a fluorescence micrograph of the capsule following live/dead cell staining. The membrane exhibits uniform thickness, well‐defined boundaries, structural integrity, and excellent sealing performance; no extracapsular fluorescently labeled cells were observed, confirming effective containment of encapsulated cells.

FIGURE 4.

(a) Schematic of cell encapsulation pouch; (b) Fluorescence micrograph of Calcein/PI dual staining in sealed capsule; (c) Proliferation kinetics of mesenchymal stem cells (MSCs) within the pouch. Number of samples (n = 5); (d) Live/dead viability assay of MSCs by fluorescence imaging; (e) Morphological characteristics of macrophages observed under optical microscopy. The order was untreated cells, cells stimulated by LPS for 48 h, and cells co‐cultured in the bag for 24–48 h after LPS stimulation.; (f) Relative mRNA expression levels of pro‐inflammatory and anti‐inflammatory markers in RAW cells following LPS stimulation. B‐Dense is a dense membrane made of commercial PU composites. B‐EM‐50 is a fiber membrane made from SPU. The significance analysis used above is a two‐way analysis of variance. The specific post‐hoc test used above is Tukey.

The cell encapsulation device requires the cells in the device to be active. In order to explore whether the prepared fibrous membrane can maintain cell activity, we tested the cell proliferation and viability kinetics of cells on different membrane substrates. Figure 4c depicts cellular proliferation and viability kinetics across different membrane matrices. Owing to the limited mass transfer capacity of the dense membrane, extensive cell death occurred in the B‐Dense group within the first day, with complete loss of viability by day 4. In contrast, BMSCs encapsulated within the B‐EM‐50 fibrous membrane exhibited sustained proliferation over 14 days, with growth dynamics comparable to those of the non‐encapsulated control group cultured on TCPS surfaces. Figure 4d displays live/dead staining outcomes at designated time points. On day 1, minimal viable cells were detected in the B‐Dense group, with total cell death evident by day 4. Conversely, both the B‐EM‐50 and transwell groups maintained high cell viability throughout the observation period. By day 7, cellular confluence exceeded 80% in both the membrane and transwell systems; by day 28, confluent monolayers with preserved metabolic activity were achieved. This is comparable to the cell proliferation and activity of a kind of amphoteric ionic polyurethane with low foreign body reaction for encapsulation device studied by Liu et al. [27].

Figure4e demonstrates morphological alterations in macrophages under varying conditions. Untreated macrophages (RAW264.7) displayed a spherical morphology, whereas LPS stimulation for 48 h induced cytoskeletal spreading. Upon co‐culture with BMSC‐encapsulated pouches, partial elongation and morphological polarization of macrophages were observed. Figure 4f shows qPCR analysis of M1/M2 macrophage polarization markers. Relative to the LPS‐only group, the BMSC co‐culture group exhibited significant upregulation of M2‐associated anti‐inflammatory genes—Arginase, IL10, and transforming growth factor‐β (TGF‐β)—with Arginase expression elevated 3.23‐fold. Concurrently, expression levels of M1‐associated pro‐inflammatory markers tumor necrosis factor‐α (TNF‐α) and iNOS were downregulated, with TNF‐α transcript levels reduced to 0.39‐fold of the LPS group. These findings suggest that encapsulated BMSCs exert immunomodulatory effects via paracrine signaling, wherein bioactive mediators diffuse through the semipermeable membrane to modulate macrophage phenotype, driving a phenotypic shift from M1 to M2 and thereby attenuating inflammatory responses.

2.3. In Vivo Experiments

The cell encapsulation device implanted in the body may be fibrotic, which affects the efficiency of material exchange and reduces the therapeutic effect of the device. Therefore, we constructed a mouse subcutaneous implantation model to explore the performance of the prepared fibrous membrane in preventing fibrosis. The subcutaneous implantation sites in rats are depicted in Figure 5a. The upper left quadrant contains EM‐0, the upper right quadrant EM‐50, and the lower region PTFE; the post‐suturing condition is illustrated in Figure 5b. At 28 days post‐implantation, varying degrees of graft folding were observed in the subcutaneous tissues. Notably, PTFE exhibited marked contraction and aggregation into compact nodular structures, whereas EM‐0 was uniformly enveloped by newly formed connective tissue. In contrast, minimal neotissue formation was observed on EM‐50, with the implanted material remaining morphologically distinct.

FIGURE 5.

In vivo biocompatibility evaluation of fibrous membranes in a rat subcutaneous implantation model. (a) Schematic representation of the surgical procedure for subcutaneous implantation of PTFE, EM‐0, and EM‐50 materials in rats; (b) Representative images of implant placement and wound closure; (c) Macroscopic appearance of retrieved circular implants following 14 days of subcutaneous residence; (d) Hematoxylin and eosin (H&E)‐stained histological sections of peri‐implant tissues at 14 and 28 days post‐implantation (asterisk denotes implant site); (e) Masson's trichrome‐stained sections of surrounding connective tissue at 14 and 28 days post‐implantation; (f) Immunofluorescence localization of inflammatory markers in periprosthetic tissue sections at 14 and 28 days (dashed white line indicates implant boundary); (g) Quantitative analysis of CD68+ macrophage infiltration in tissue sections at 14 and 28 days after subcutaneous implantation. Number of samples (n = 3). The significance analysis used above is a two‐way analysis of variance. The specific post‐hoc test used above is Tukey.

Histological analysis via H&E staining is presented in Figure 5c. At 14 days, the PTFE membrane displayed significant shrinkage and partial separation from adjacent host tissue, resulting in cavity formation, with only sparse cellular attachment on its surface. H&E staining (Figure 5d) reveals that both EM‐0 and EM‐50 were covered by a continuous layer of host‐derived tissue, with greater thickness observed in the perigraft capsule surrounding EM‐0 compared to EM‐50. By day 28, periprosthetic tissue distribution became more homogeneous across all groups. Adhesion between PTFE and surrounding tissues improved, accompanied by increased thickness and cellular density in the fibrous capsule around EM‐0. In contrast, the tissue interface surrounding EM‐50 remained loosely organized.

Masson's trichrome staining (Figure 5e) reveals collagen deposition and myofibroblastic differentiation in the periprosthetic milieu. At 14 days, muscle fiber‐like, eosinophilic structures were evident around PTFE and EM‐0, indicative of early myofibroblast infiltration or skeletal muscle alignment, while no such structures were apparent around EM‐50. By day 28, these fiber‐like elements had regressed in the PTFE group, whereas EM‐0 maintained a dense network of aligned muscle fiber‐like tissue. No evidence of myogenic differentiation or organized fiber formation was detected around EM‐50 at this time point.

Immunofluorescence staining for macrophages is shown in Figure 5f. At 14 days post‐implantation, robust CD68+ macrophage infiltration was observed in the perigraft regions of PTFE and EM‐0, whereas EM‐50 exhibited minimal CD68+ signal. At 28 days, further accumulation of CD68+ cells was noted around PTFE and EM‐0, consistent with sustained inflammatory engagement. In contrast, EM‐50 continued to show limited macrophage recruitment. Semi‐quantitative analysis of CD68+ expression (Figure 5g) confirmed progressive macrophage infiltration in PTFE and EM‐0 groups from 14 to 28 days, while EM‐50 maintained consistently low levels of macrophage presence. Compared with the in vitro compatibility test of zwitterionic polyurethane studied by Wang et al., the modified polyurethane fiber membrane in this paper has better in vivo compatibility test and better biocompatibility [28].

3. Discussion

The block copolymer materials synthesized in this study exhibit proper mechanical properties, fulfilling the structural integrity requirements for practical biomedical applications. Through rational molecular design incorporating hydrophilic segments, the surface hydrophilicity of the material was markedly improved, thereby enhancing the bidirectional diffusion of nutrients, metabolites, and signaling molecules across the membrane. This facilitated a favorable microenvironment conducive to prolonged cell viability and functional stability.

Experimental results demonstrate that the modified fibrous membrane significantly suppresses non‐specific protein adsorption owing to enhanced hydrophilicity. The reduction in transmembrane protein deposition not only preserves the patency of the pore architecture and ensures sustained transport efficiency but also mitigates chronic inflammatory responses associated with foreign body reactions, thus minimizing fibrotic encapsulation. Furthermore, morphological analysis reveals that fiber diameter decreases with increasing SPU (zwitterionic polyurethane) content. This phenomenon is attributed to the incorporation of zwitterionic moieties, which elevate the overall charge density of the polymer system. During electrospinning, the intensified Coulombic forces under a high‐voltage electric field lead to enhanced jet acceleration and pronounced axial stretching, ultimately yielding nanofibers with reduced diameters. From a functional perspective, smaller fiber dimensions correspond to diminished pore sizes, which effectively prevent cellular leakage while maintaining adequate permeability, thereby improving the safety and reliability of the encapsulation system.

The encapsulation device based on the engineered fibrous membrane achieved efficient cell immobilization. In vitro studies confirmed that optimal porosity allowed for sustained cell survival and preserved secretory activity within the device. Moreover, secreted bioactive factors were able to diffuse into surrounding tissues and elicit measurable physiological responses, indicating strong potential for therapeutic applications.

In vivo implantation experiments revealed perimaterial collagenous tissue formation across all groups. However, immunofluorescence analysis showed minimal infiltration of CD68+ macrophages around the modified polyurethane in the EM‐50 group, indicating potent inhibition of macrophage activation in vivo. Additionally, Masson's trichrome staining detected no muscle fiber ingrowth adjacent to the implanted EM‐50 membranes, suggesting that observed collagen deposition likely originated from host‐derived extracellular matrix remodeling rather than pathological fibrosis. In contrast, PTFE and unmodified EM‐0 materials exhibited evident muscle fiber integration, presumably due to their highly hydrophobic surfaces promoting nonspecific protein adsorption, which triggers macrophage activation, initiates inflammatory cascades, and promotes cellular recruitment and proliferation—culminating in dense fibrous encapsulation.

Compared to alternative cell encapsulation platforms such as microcapsules, hydrogel entrapment, or multilayer thin‐film systems, the fibrous membrane‐based bag structure offers distinct advantages in terms of mechanical robustness, tunable permeability, and scalability. Unlike mechanically fragile microcapsules, the fibrous mat provides structural resilience; in comparison to hydrogels with inherent diffusional barriers, the porous fibrous network enables more efficient macromolecular transport. Nevertheless, potential limitations remain, including possible degradation‐induced alterations in membrane porosity and barrier function following long‐term implantation. Additionally, current designs lack optimized local immunomodulatory functionality.

Several limitations exist in the present study. Notably, the degradation behavior of the material system remains insufficiently characterized, and data regarding the temporal evolution of structural parameters—including pore size, porosity, and mechanical integrity in vivo—are limited. Moreover, the mouse subcutaneous implantation model in animal experiments has certain limitations, and it is impossible to reproduce the actual microenvironment of the cell encapsulation device at other targeted sites such as the abdomen. In future work, the specific diseases of the cell encapsulation device will be studied, the dynamic changes of key characteristics after implantation will be systematically studied, and the mechanism relationship between material stability and host immune response will be elucidated. Such insights will provide a foundational framework for the rational design of next‐generation biocompatible encapsulation materials with enhanced performance and clinical translatability.

4. Conclusion

In this study, an amphiphilic polyurethane‐based block copolymer incorporating hydrophilic zwitterionic segments was designed and synthesized via a coupling reaction between hydroxyl‐terminated sulfobetaine methacrylate (SBMA) and isocyanate‐terminated polyurethane prepolymers. The incorporation of hydrophilic SBMA moieties markedly enhanced surface hydrophilicity, thereby reducing protein fouling through hydration layer formation. Nanofibrous membranes with tunable copolymer compositions were fabricated by electrospinning, yielding a well‐defined 3D porous architecture with submicron fiber diameters and controlled pore dimensions. This structural configuration facilitates cellular adhesion and proliferation while minimizing non‐specific protein adsorption, demonstrating superior biocompatibility in vitro. Encapsulation of bone marrow‐derived mesenchymal stem cells (BMSCs) within EM‐50 membranes effectively prevented cell leakage and permitted unrestricted diffusion of nutrients and signaling molecules, enabling sustained cell viability and stimulus‐responsive exosome secretion. In vivo implantation studies revealed no significant acute or chronic inflammatory responses at 14 and 28 days post‐implantation, with marked suppression of fibrous capsule formation, indicating effective mitigation of host immune recognition. The zwitterionic surface‐modified polyurethane nanofibrous membrane exhibits excellent hemocompatibility, low immunogenicity, and robust immune‐isolation capability, positioning it as a promising candidate for next‐generation cell encapsulation platforms in regenerative medicine and therapeutic cell delivery applications.

5. Experimental Section

5.1. Materials

The N‐(3‐Sulfopropyl)‐N‐methacryloxyethyl‐N,N‐dimethylammonium betaine, 2,2'‐azobis(2‐methyl‐N‐(2‐hydroxyethyl)propionamide), N, N‐dimethylformamide, Dimethyl sulfoxide (DMSO), Hexafluoroisopropanol, Lipopolysaccharide (LPS), Lysozyme, Bovine albumin, and the catalyst Bicat8108 were purchased from a leading US chemical company. The polybutylene glycol was purchased from the Sigma–Aldrich company. The hexamethylene diisocyanate (HDI) was purchased from Tokyo Chemical Industry. The fetal bovine serum, DMEM, and PBS were purchased from the Gibco Life Sciences company. The Cell Counting Kit‐8 (CCK‐8) and calcein‐AM/PI double staining kit were purchased from Dojindo Laboratories.

5.2. Synthesis of the Block Copolymer PSBMA‐PU‐PSBMA

5.2.1. Synthesis of PSBMA‐OH

Dissolve 4.716 g of N‐(3‐Sulfopropyl)‐N‐methacryloxyethyl‐N,N‐dimethylammonium betaine in 20 mL of dimethyl sulfoxide and 0.48 g of 2,2'‐azobis(2‐methyl‐N‐(2‐hydroxyethyl)propionamide) as the initiator was then added. Nitrogen was introduced into the system for 30 min. Polymerization was conducted under a nitrogen atmosphere via ultraviolet light (365 nm) irradiation for 1 h to obtain PSBMA‐OH bearing monohydroxy end groups.

5.2.2. Synthesis of Polyurethane Prepolymers

9 mmol polybutylene glycol (Mn 250) was dehydrated at 100°C under vacuum for 4 h, then cooled to room temperature Subsequently, 10 mmol hexamethylene diisocyanate (HDI), 3 wt.‰Bicat8108 catalyst, and 2 mL of anhydrous N, N‐dimethylformamide (DMF) were added. The reaction mixture was heated to 80°C and stirred for 1 h under a nitrogen atmosphere, yielding a diisocyanate‐terminated prepolymer denoted as pPU.

5.2.3. Synthesis of Block Copolymers

PSBMA‐OH was reacted with pPU at 65°C for 2 h to synthesize corresponding block copolymers PSBMA‐PU‐PSBMA. The product solution after the reaction was poured into a large amount of deionized water to collect the polymer. These resulting copolymers were dried in a vacuum oven at 40°C for 24 h and designated as SPU.

5.3. Preparation of Electrospinning Fiber Membrane

The synthesized block copolymer was dissolved at room temperature in hexafluoroisopropanol (HFIP) together with commercial polyurethane (PU) at varying weight ratios. A 4 mL aliquot of the homogeneous polymer solution was loaded into a 5 mL plastic syringe equipped with a 23 G blunt‐tip needle serving as the spinneret. Electrospinning was performed under a constant voltage of 10 kV, with a tip‐to‐collector distance of 15 cm. The solution was fed at a flow rate of 1.5 mL/h using a syringe pump. Fibers were collected on a grounded aluminum plate positioned in the path of the charged jet. After electrospinning, the resulting nanofibrous membrane was immersed in deionized water for 24 h to remove residual solvent. The membrane was then carefully peeled from the collector and dried under vacuum at 40°C to constant weight. Membranes containing 0%, 20%, 35%, and 50% SPU by weight were coded as EM‐0, EM‐20, EM‐35, and EM‐50, respectively.

5.4. Characterization

5.4.1. Chemical Structures Characterization

Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy (ATR‐FTIR, Vertex 70, Bruker Optics, Co, USA) was utilized to characterize the polymeric backbone and functional group composition, while X‐ray Photoelectron Spectroscopy (XPS, K‐Alpha, Thermo Fisher Scientific, Co, USA) was conducted to determine the surface elemental composition and atomic ratios of C, N, O, and S in the EM‐50 fiber membrane. The H nuclear magnetic resonance (400 M, Bruker) hydrogen spectrum can analyze the hydrogen atoms in different chemical environments to infer the structure of the sample. The sample was dissolved in 600 µl D₂O and analyzed at 25°C. The scan range was 3–16 ppm, and the scan was 32 cycles.

5.4.2. Gel Permeation Chromatography (GPC)

The molecular weight and distribution of the polymer were determined on a Viscotek GPC instrument. The sample was dissolved in chromatographic grade N, N‐dimethylformamide at a concentration of 10 mg/mL. The detector was a Viscotek TDA 305 triple detection system, the chromatographic column was a CLM30006T3000M chromatographic column, the mobile phase was chromatographic grade N, N‐dimethylformamide, the flow rate was set to 1 mL/min, the column temperature was 60 C, and the standard sample was polystyrene.

5.4.3. Calculation of Porosity

Using n‐butanol as a wetting agent, the membrane porosity of a circular membrane sample with a diameter of 1 cm was measured. The samples were immersed in a n‐butanol bath for 24 h. Before soaking, the samples were dried and weighed (M₀). After soaking, the surface liquid was removed with non‐woven fabric, and the sample (M_B) was weighed again. Then the porosity P (%) of the sample was determined using the following formula, where ρ_B and ρ₀ represent the density of butanol and polymer, respectively.

$$\mathrm{P}=\frac{\frac{{\mathrm{M}}_{\mathrm{B}}-{\mathrm{M}}_0}{{\rho }_{\mathrm{B}}}}{\frac{{\mathrm{M}}_0}{{\rho }_0}+\frac{{\mathrm{M}}_{\mathrm{B}}-{\mathrm{M}}_0}{{\rho }_{\mathrm{B}}}}$$ (1)

5.4.4. Physical Properties Evaluation

Physical properties evaluation. Thermogravimetric analysis (TGA) was performed to evaluate the thermal stability of the block copolymer. The hydrophilicity of block copolymers, porous membranes, and fiber membranes was assessed via water contact angle measurements. Mechanical properties of the fiber membranes were determined using a universal testing machine under uniaxial tension. Surface morphology was examined by Scanning Electron Microscopy (SEM, Merlin, ZEISS, Germany) at an accelerating voltage of 5 kV.

5.4.5. Membrane Water Flux

Membrane water flux was measured using a cross‐flow filtration setup. A peristaltic pump ensured a constant feed flow, with the porous membrane sealed between the filtration cell components. By adjusting the pump flow rate and valve resistance, the transmembrane pressure was stabilized at 0.1 mPa. Water flux was calculated based on the volume of permeate collected over a defined time interval normalized by the effective membrane area.

5.4.6. Protein Adsorption on Membrane

The BCA assay was utilized to determine the extent of non‐specific protein adsorption on material surfaces in an indirect manner. A 2 mg/mL bovine albumin (BSA) solution was dispensed dropwise onto the fiber membrane surfaces. Samples were incubated under agitation at 37°C for 24 h. Each experimental group included a blank control (defined as 100% adsorption) and three replicates. Aliquots of 20 µL from the incubation supernatant were mixed with 200 µL of BCA working reagent and incubated at 37°C for 30 min. Absorbance was measured at 562 nm, and residual BSA concentrations were quantified via a calibration curve.

5.5. Cell Experiment

5.5.1. Cell Culture

Mouse fibroblasts (L929) and macrophages (RAW264.7) were cultured in high‐glucose DMEM (C11995500BT, Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS, 10099141C, Gibco) under standard conditions: 37°C, 5% CO₂, and 95% relative humidity. Cells exhibiting ≥ 80% confluence after 72 h were selected for subculture. Adherent cells were dissociated using 1% (w/v) trypsin‐EDTA solution for passaging. Experiments were conducted at the fifth passage (P5). For in vitro polarization, RAW264.7 macrophages were stimulated with 1200 ng/mL lipopolysaccharide (LPS) for 24 h to induce pro‐inflammatory activation.

5.5.2. Cell Encapsulation Experiments

MSCs are loaded in the sacs by injection to evaluate their capacity to retain cells and enable proliferation. To evaluate the permeable properties of the membrane, the cells sealed in sacs were cocultured with RAW cells induced by LPS. The characteristic gene of macrophage phenotype was tested by qPCR.

5.5.3. qPCR Experiments

RAW264.7 cells were cultured and lysed at predetermined days. Total ribonucleic acid (RNA) was extracted using TRIzol reagent. Following RNA isolation, DNase I treatment was applied to remove any contaminating deoxyriboNucleic acid (DNA). The con­centration and purity of RNA were analyzed using a Nanodrop spec­trophotometer (Thermo). cDNA was synthesized using the HiFiScript cDNA Synthesis Kit and RNA reverse transcription system (TC‐412, TECHNE). Quantitative reverse transcriptase‐polymerase chain reaction (qRT‐PCR) was performed using an Applied Biosystems QuantStudio 6 Flex Real‐Time PCR System (Thermo) with UltraSYBR Mixture. The sequences of the forward primers and reverse primers are listed in Table 1. Relative mRNA levels were calculated and normalized to GAPDH expression levels. All qRT‐PCR experiments were performed in triplicate to ensure reproducibility.

TABLE 1.

Primer sequences for RAW264.7 cells.

Gene of RAW264.7	Primer sequences
GAPDH	Forward5’‐GTCGGTGTGAACGGATTTGG‐3’ Reverse5’‐CTGGAAGATGGTGATGGGCT‐3’
TNF‐α	Forward5’‐CAACGGCATGGATCTCAAAG‐3’ Reverse 5’‐ CAGCCTTGTCCCTTGAAGA ‐3’
iNOS	Forward5’‐ATGCTACTGGAGGTGGGTG ‐3’ Reverse5’‐GATGTTGTAGCGCTGTGTGT‐3’
IL‐10	Forward5’‐GGTGAGAAGCTGAAGACCCT‐3’ Reverse5’‐CACTCTTCACCTGCTCCACTG‐3’
Arginase	Forward5’‐CATATCTGCCAAAGACATCG‐3’ Reverse5’‐GGTCTCTTCCATCACCTTGC ‐3’
TGF‐β1	Forward5’‐TGGAGCAACATGTGGAACTC‐3’ Reverse5’‐TGCCGTACAACTCCAGTGAC ‐3’

5.6. Animal Experimental Studies

The experimental model utilized subcutaneous implantation in rats. Circular specimens of unmodified polyurethane fiber membrane (EM‐0) and surface‐modified fiber membrane (EM‐50), each with a diameter of 10 mm, constituted the experimental groups, whereas polytetrafluoroethylene (PTFE) porous membrane served as the control. The foreign body response (FBR) to EM‐0 and EM‐50 was assessed at two time points post‐implantation: 14 and 28 days. As a control, a commercially available PTFE porous membrane with a nominal pore size of 0.22 µm, sourced from a sterile Jet filter, was employed. Six healthy male Sprague‐Dawley rats were acclimated under controlled environmental conditions (22°C, 50% relative humidity) with a 12 h light/dark cycle and libitum access to food and water. Prior to surgery, each rat was anesthetized, the dorsal region was shaved and disinfected with 70% ethanol, and three parallel 10 mm incisions were made bilaterally along the dorsum, approximately 2 cm lateral to the midline and cranial to the caudal vertebrae. Sterilized circular membranes (10 mm diameter) were implanted subcutaneously into each incision site. The control group received PTFE membranes, while the experimental groups received either EM‐0 or EM‐50 electrospun membranes. Postoperative prophylactic antibiotics were administered to mitigate infection risk. At day 14, three randomly selected animals were euthanized. Dorsal skin sections containing the implant and surrounding tissue were excised, then processed for histological analysis via fixation (in 10% neutral buffered formalin), dehydration through graded ethanol series, paraffin embedding, microtome sectioning, and staining with hematoxylin and eosin (H&E), Masson's trichrome, and immunohistochemical markers. The remaining three animals were similarly euthanized at day 28, and identical tissue harvesting and histological processing procedures were performed. All animal experiments have been approved by the Jinan University Laboratory Animal Welfare and Ethics Committee (No. IACUC‐20240116‐04) and conducted by following the guidelines from the Ethics Committee and national regulations of China.

5.7. Statistical Analysis

All statistical analyses were performed using the Prism 9.0 software (GraphPad Software, Inc., SanDiego, CA). Statistical significance was determined using one‐ or two‐way ANOVA, with significance levels set at p < 0.05 (^*), p < 0.01 (^**), p and < 0.001 (^***), respectively.

Funding

Guangdong Basic and Applied Basic Research Foundation (2023A1515012035), Guangdong Key Areas Research and Development Programs (2025B1111090001).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.