Cross-shaped nanoporous poly(methyl methacrylate) fibers for selective adsorption of β 2 -microglobulin and other middle-molecular-weight proteins

Abstract

Background:

Dialysis-related amyloidosis (DRA) is a serious complication in chronic kidney disease patients on long-term dialysis, caused by β₂-microglobulin (β₂-MG) accumulation, and remains challenging owing to the growing dialysis population and extended treatment duration.

Objectives:

To develop and evaluate stereocomplex poly(methyl methacrylate) (PMMA) adsorbent fibers with optimized cross-sectional and nanoporous structures for efficient, selective β₂-MG removal.

Methods:

Structured PMMA fibers were fabricated via dry-wet spinning. The adsorption performance of optimized fibers was evaluated with serum containing β₂-MG and other solutes, using scanning electron microscopy (SEM), three-dimensional transmission electron microscopy (3D-TEM), and time-of-flight secondary ion mass spectrometry (TOF-SIMS). Pressure loss simulations compared fiber- and bead-packed columns.

Results:

Fibers with 12–15 nm pores exhibited maximal β₂-MG adsorption. Cross-shaped fibers exhibited the highest blood-contact surface area and adsorption capacity. A thin surface-dense layer (<0.1 μm) improved β₂-MG diffusion while limiting albumin uptake. TOF-SIMS confirmed uniform β₂-MG penetration, with albumin confined near the fiber surface. Adsorption exceeded 50% for mid-weight proteins (~52 kDa), including IL-6, α₁-microglobulin (α₁-MG), and TNF-α. Pressure loss simulations showed that fiber-packed columns had lower resistance than bead-packed columns.

Conclusions:

Optimized cross-shaped PMMA fibers enable efficient, selective β₂-MG removal and favorable flow dynamics for hemoperfusion therapies targeting mid-weight uremic toxins.

Introduction

Various diseases, including chronic diseases such as dialysis-related complications and certain autoimmune disorders, and acute diseases such as sepsis, are caused by the accumulation of pathogenic proteins in the blood. One treatment approach for these diseases involves removing pathogenic proteins using adsorption columns. Direct elimination of these pathogenic proteins has the potential to serve as a fundamental treatment approach.

Dialysis-related complications occur in patients with chronic kidney disease (CKD) undergoing dialysis. They are primarily caused by the accumulation of uremic toxins in the blood and body during long-term dialysis. Among these toxins, β₂-microglobulin (β₂-MG) is known to increase in concentration by 10 to 50 times compared with that in healthy individuals and has been identified as a pathogenic protein responsible for dialysis-related amyloidosis (DRA). DRA is a form of systemic amyloidosis in which β₂-MG forms high-molecular-weight aggregates and is deposited in bones and joints, often leading to carpal tunnel syndrome or dialysis-related spondyloarthropathy. ¹

Deposits have also been reported in the heart and gastrointestinal tract, ² which not only reduce the quality of life and activities of daily living but also negatively affect life prognosis. ³

Currently, there are no established pharmacological treatments for DRA. The therapeutic approach is to actively remove β₂-MG during dialysis therapy to prevent onset or slow disease progression. In particular, for post-onset management, eliminating β₂-MG together with water, urea, and other low-molecular-weight compounds using a dialyzer alone is insufficient; therefore, combining this with a β₂-MG adsorption column to achieve more aggressive removal is also considered. ⁴

Since 1977, dialyzers with poly(methyl methacrylate) (PMMA) membranes have been used clinically in patients with CKD. ⁵ These dialyzers not only perform basic filtration but also exhibit adsorption capabilities, enabling active removal of middle-molecular-weight proteins that are difficult to eliminate by filtration alone. ⁶ This adsorption property enhances β₂-MG removal compared with dialyzers using cellulose membranes, thereby reducing the risk of DRA. ⁷ The PMMA membrane used in dialyzers is a stereocomplex formed by blending isotactic and syndiotactic PMMA, resulting in a unique supramolecular structure. ⁸ Compared with isotactic or syndiotactic PMMA alone, stereocomplex PMMA exhibits superior blood compatibility and higher saturated protein adsorption capacity.^9,10

In this study, we developed a β₂-MG adsorbent using PMMA. The design of the adsorbent shape and internal nanoporous structure is a critical factor for enhancing the adsorption performance. When used in adsorption columns, the shape of the adsorbent influences both the blood-contact surface area and blood flow. Increasing the contact surface area can improve the protein adsorption rate; however, it may also lead to increased pressure loss within the column, potentially causing risks such as hemolysis. Bead-shaped adsorbents are commonly used in commercially available columns. However, the flow paths between the beads tend to be tortuous, resulting in significant pressure losses. To address this issue, we were inspired by the concept of creating a linear blood flow and attempted to develop long fiber-shaped adsorbents that could be arranged parallel to the flow path. Furthermore, we modified the cross-sectional shape of the fibers to increase their blood-contact surface area. In designing a nanoporous structure, it is essential to optimize the nanopore diameter for efficient protein adsorption and to ensure a continuous, interconnected porous structure extending from the surface to the interior of the adsorbent fibers, so that the entire fiber contributes to the adsorption capacity.

Materials and methods

Preparation of adsorbent fibers

As previously reported, ⁵ a spinning dope containing a mixture of isotactic and syndiotactic PMMA was extruded through a spinneret. Stereocomplex PMMA structures were formed in the dry-spinning zone via phase separation and growth. The resulting fibers were solidified in a coagulation bath, washed, and wound onto a bobbin to obtain PMMA fibers. The spinneret nozzle was circular, elliptical, Y-shaped, or cross-shaped, depending on the desired cross-sectional geometry. The degree of fiber modification was calculated using the following equation:

$$Modificationdegree=\frac{\begin{array}{l}Perimeterof\\ fibercross-section\end{array}}{\begin{array}{l}Perimeterofacircularfiberwith\\ equivalentcross-sectionalarea\end{array}}$$

Measurement of β₂-MG adsorption amount

Fibers (0.1 cm³) were immersed in 12 mL of fetal bovine serum (FBS; Biowest SAS) containing 35 mg/L human β₂-MG (FUJIFILM Wako Pure Chemical Corporation) and shaken at 37°C for 1 h. β₂-MG concentrations before and after immersion were determined by immunoprecipitation at the SRL, and the adsorption capacity was calculated per unit fiber volume.

Measurement of the adsorption rate

Fibers (0.2 cm³) were immersed in 3 mL of FBS containing various solutes: β₂-MG and α₁-microglobulin (α₁-MG) (both concentrated from peritoneal dialysis fluid); IL-8 and IL-10 (FUJIFILM Wako Pure Chemical Corporation); IL-6 (Kamakura Techno Science); TNF-α (PeproTech); and MMP-3 (R&D Systems). The mixture was then shaken at 37°C for 2 h. The initial solute concentrations were as follows: β₂-MG, 35 mg/L; α₁-MG, 35 mg/L; IL-8, 7 μg/L; IL-10, 500 ng/L; IL-6, 50 ng/L; TNF-α, 14 μg/L; and MMP-3, 680 μg/L. Solute concentrations were determined using the following methods: ELISA was used for IL-8, IL-10, and TNF-α (R&D Systems), ELISA was used for IL-6 (Kamakura Techno Science), and the BCG method was used for albumin (FUJIFILM Wako Pure Chemical Corporation). Other solutes were quantified by SRL, Inc. using a latex agglutination immunoassay for β₂-MG and α₁-MG, and a latex turbidimetric immunoassay for MMP-3. Similarly, the fibers were immersed in human serum (Cosmo Bio), and the adsorption rates of the solutes were measured. Solute concentrations in these samples were measured using the following methods: liquid chromatography–mass spectrometry for total amino acids, chemiluminescent enzyme immunoassay for Vitamin B₁₂, and turbidimetric immunoassays for immunoglobulins (IgA, IgG, and IgM) and complement components (C3 and C4). All measurements were outsourced to SRL Inc.

Scanning electron microscopy (SEM)

The fibers were frozen by immersion in liquid nitrogen, folded to expose the cross section, and vacuum dried. A thin Pt film was sputter-coated onto the sample surface for conductivity. The samples were observed using SEM (SEM-EDX Type H, Hitachi High-Tech).

Three-dimensional transmission electron microscopy (3D-TEM)

The dried fiber samples were embedded in resin to form blocks. After electron staining, ultrathin sections were prepared using an ultramicrotome (EM UC7; Leica) and mounted on colloidal-coated TEM grids. TEM was performed using a JEM-F200 instrument (JEOL) operated at 200 kV. Tilt-series TEM images were collected from −70° to +70° in 2° increments. 3D reconstruction was performed using simultaneous iterative reconstruction implemented in Composer (System in Frontier). Image alignment, grayscale correction, and binarization were conducted using Avizo software (Thermo Fisher Scientific), and a pore network model was generated using the Avizo pore network modeling extension.

Time-of-flight-secondary ion mass spectrometry (TOF-SIMS)

The fibers were immersed in phosphate-buffered saline (FUJIFILM Wako Pure Chemical Corporation) containing either human β₂-MG (40 mg/L, FUJIFILM Wako Pure Chemical Corporation) or human serum albumin (40,000 mg/L, Sigma-Aldrich) and shaken at 37°C for 24 h. After shaking, the fibers were embedded in resin, sectioned using an ultramicrotome (Leica, EM UC), and dried. Cross-sectional TOF-SIMS images were acquired using a TOF-SIMS 5 instrument (IONTOF GmbH) equipped with a 60 kV Bi₃²⁺ primary ion beam and a low-energy electron beam for charge compensation. Measurements were performed in the fast imaging mode with a mass resolution of ~300 m/Δm and a spatial resolution of ~300 nm. Characteristic secondary ions analyzed included C₄H₅O₂⁻ for PMMA, CN⁻ and CNO⁻ for proteins, and C₂H₃O⁻ for resins. Secondary ion images were normalized to the total ion intensity, and line profiles of normalized intensity (from the fiber center to edge) were used to quantify the protein adsorption distribution. Image processing and visualization were performed using SurfaceLab 7.0 (IONTOF).

Pressure loss simulation

Pressure loss during blood flow was simulated for columns packed with either cross-shaped PMMA fibers or spherical beads as adsorbents. Each column had an inner diameter ( $D_c$ ) of 65 mm, a length ( $L$ ) of 67 mm, and a void fraction ( $\epsilon$ ) of 0.35.

To calculate the pressure loss in a column packed with cross-shaped fibers, a modified Hagen–Poiseuille equation ¹¹ was used. The modified Hagen–Poiseuille equation can be written as:

$$\begin{array}{l}Pressurelossincross-shapedfiberspacked\\ column(\Delta P_{cf})=32\left(\frac{\mu Lu}{{d_{cf}}^2}\right)\end{array}$$

where $\mu$ is the viscosity of blood (3 cP), $u$ is the average linear velocity of blood within the column, and $d_{cf}$ is the hydraulic equivalent diameter of the flow path, calculated as:

$$d_{cf}=4\left(\frac{\begin{array}{l}Flowchannelcross\\ -sectionalareainsidethecolumn\end{array}}{Wettedperimeter}\right)$$

The wetted perimeter is given by:

$$Wettedperimeter=D_c\pi +C\pi N$$

where $C$ is the perimeter of the fiber cross section and $N$ is the number of fibers.

To calculate the pressure loss in a column packed with bead-shaped fibers, the Ergun equation ¹² was used. The Ergun equation can be written as:

$$\begin{array}{l}Pressurelossinbeadspackedcolumn\Delta P_{be}\\ =\left\{150\left(\frac{\left(1-\epsilon \right)}{R{e}_{be}}\right)+1.75\right\}\frac{1-\epsilon }{{\epsilon }^3}*\frac{\rho v^{2}L}{D_{be}}\end{array}$$

where $\rho$ is the density of blood (1050 kg/m³), $D_{be}$ is the bead diameter, and $v$ is the superficial velocity of blood in an empty column.

The Reynolds number ( $R{e}_{be}$ ) was calculated using the following equation:

$$R{e}_{be}=\left(\frac{D_{be}v\rho }{\mu }\right)$$

Results

We first investigated the optimal nanopore diameter in circular PMMA fibers for β₂-MG adsorption. Fibers with average nanopore diameters ranging from 10 to 18 nm were evaluated, and the results indicated that the amount of adsorbed β₂-MG was maximized within the 12–15 nm range (Figure 1).

Figure 1.

β₂-MG adsorption amount as a function of average nanopore diameter of circular-shaped PMMA fibers.

Based on these findings, the average nanopore diameter was set to ~14 nm for subsequent experiments. Subsequently, we examined the effect of the cross-sectional shape of the fibers on the blood-contact surface area. Various fiber geometries, including circular, elliptical, Y-shaped, and cross-shaped, were fabricated and characterized (Table 1). Although the cross-sectional areas of all fibers were nearly identical, the modification degrees were 1.1 for elliptical, 1.2 for Y-shaped, and 1.3 for cross-shaped fibers, respectively. The cross-shaped fibers exhibited the highest degree of modification, and the β₂-MG adsorption increased proportionally with this degree. These findings suggest that increasing the blood-contact surface area through cross-sectional modification of the adsorbent fibers improves the adsorption performance.

Table 1.

Evaluation of PMMA fibers with various cross-sectional shapes.

Cross-sectional shape	Circular	Elliptical	Y-shaped	Cross-shaped
Example of cross-sectional image
Cross-sectional area (mm2)	0.0106	0.0100	0.0104	0.0102
(1) Average perimeter of fiber cross-section (µm)	368	410	449	480
(2) Perimeter of a circular fiber with equivalent cross-sectional area (µm)	364	354	361	358
(1)/(2) Modification degree (–)	1.0	1.1	1.2	1.3
β2-MG adsorption amount (mg/cm3)	0.27	0.31	0.33	0.35

Observation of the surface region of the cross section of the cross-shaped PMMA fibers revealed a dense outer layer (Figure 2(A)). The nanopores in this layer were smaller than those in the fiber interior and could not be clearly visualized using SEM. This dense layer was formed at the air–liquid interface when the spinning dope was extruded from the spinneret during fiber fabrication. Although such a layer is expected to exhibit size-selective solute separation, an excessively thick dense layer may hinder the diffusion of solutes, such as β₂-MG, into the fiber interior. To evaluate this effect, fibers with different dense layer thicknesses were prepared, and their β₂-MG adsorption capacities were measured. As shown in Figure 2(B), reducing the dense layer thickness to below 0.1 μm improved the adsorption performance. SEM images of cross-shaped PMMA fibers with dense layer thickness below 0.1 μm are shown in Figure 2(C) to (F). These images indicate that apart from the outer dense layer, the fiber structure was homogeneous, suggesting that the entire fiber served as an adsorption site.

Figure 2.

Evaluation of the surface-dense layer of PMMA cross-shaped fibers: (A) SEM image of the dense layer in the fiber cross-section and (B) relationship between dense layer thickness and β₂-MG adsorption amount. SEM images of the fiber cross-section: (C) overall image and (D, E, and F) magnified views of the central, intermediate, and near-surface regions, respectively. 3D-TEM images of the outer surface of a fiber cross-section: (G) 3D-TEM image, (H) 3D-reconstruction, and (I) pore network model. Images (H) and (I) were obtained through additional analysis of (G). (J) shows the relative abundance of PMMA and pores corresponding to (H).

Figure 2(G) shows a section of the 3D-TEM image of the near-surface region of the fiber cross section, Figure 2(H) presents the 3D reconstruction, and Figure 2(I) illustrates a pore network model. Although a dense layer was present on the outer surface, the nanopore structure was continuous throughout the interior, forming a network-like connection. This continuity suggests that blood proteins can diffuse through the fibers and reach the central region. Furthermore, Figure 2(J) shows that within the 45 nm-thick dense layers, the PMMA fraction was higher than the pore fraction, whereas the inner regions contained more pores than PMMA. The average nanopore diameters in the dense and inner layers obtained from 3D-TEM analysis were 9 and 16 nm, respectively (data not shown).

To evaluate the adsorption selectivity of the cross-shaped PMMA fibers, we measured their adsorption rates using serum samples containing various solutes. Figure 3 shows the relationship between the molecular weight of each solute and its adsorption rate. Adsorption was the highest for β₂-MG (molecular weight: 12 kDa). Other middle-molecular-weight proteins, such as IL-6 (25 kDa), α₁-MG (33 kDa), and TNF-α (52 kDa), also exhibited high adsorption rates (>50%), whereas large-molecule proteins, such as albumin (66 kDa), antibodies, and complement (⩾150 kDa), exhibited low adsorption rates (<20%). Small molecules such as amino acids and vitamin E also exhibit low adsorption rates.

Figure 3.

Adsorption rates of various proteins on PMMA cross-shaped fibers. Each value represents the average of three independent measurements, and the error bars indicate the standard deviation.

To investigate the adsorption distribution of proteins within the cross-shaped PMMA fibers, TOF-SIMS mapping was performed on cross sections of the fibers after exposure to either β₂-MG (40 mg/L) or albumin (40,000 mg/L) aqueous solutions. The SIMS data are presented in Figure 4(A) to (C), and the line profiles obtained along the white dashed lines in Figure 4(A) to (C) are presented in Figure 4(D). The results indicated that adsorption occurred uniformly around the fiber (360°) and that β₂-MG diffused into the interior of the fiber during adsorption. In contrast, albumin adsorption was primarily localized near the fiber surface, despite its 1000 times higher concentration, possibly owing to its large molecular weight (albumin is 66 kDa, β₂-MG is 12 kDa.) and limited diffusion through the dense surface layer.

Figure 4.

TOF-SIMS mapping images of adsorbed proteins on PMMA fiber cross sections after 24 h of adsorption: (A) before adsorption, (B) after contact with 40 mg/L β₂-MGl, and (C) after contact with 40,000 mg/L albumin. In the overlay images, red indicates PMMA and green indicates the presence of proteins. (D) Line profiles along the white dashed lines in (B) and (C).

Finally, the pressure loss during hemoperfusion was simulated by packing the developed cross-shaped PMMA fibers into a column and comparing the results with those obtained using bead-shaped adsorbents. As shown in Figure 5, when the equivalent circular diameter of the fibers or beads was reduced from 700 μm to 114 μm, the total surface area of the adsorbent in the column increased, and the pressure loss with the fibers remained lower than that observed with the beads.

Figure 5.

Simulation of pressure loss during whole-blood perfusion when the adsorbent was packed in a column as cross-shaped fibers or beads. The column has an inner diameter of 65 mm, a length of 67 mm, and an open-space ratio of 0.35. The data points, from left to right, correspond to calculated results for equivalent circular diameter of the cross-shaped fibers or bead diameters of 700, 400, 275, 200, 160, 130, and 114 μm.

Discussion

The incidence of DRA, along with its major clinical manifestation, carpal tunnel syndrome, has declined in recent years owing to advancements in dialysis therapy. However, given the global increase in dialysis patients and the prolonged duration of treatment, DRA is expected to remain a significant therapeutic challenge in the future. ¹³

In this study, we developed an adsorbent for β₂-MG, a pathogenic protein responsible for DRA. Traditionally, hemoperfusion columns employ bead-shaped adsorbents, which create complex and tortuous blood flow paths. ¹⁴ In contrast, the cross-shaped straight fibers developed in this study can be aligned parallel to the blood flow, enabling a linear flow path within the column. As shown in Figure 5, even with a fiber diameter of approximately 114 μm (narrower than conventional hollow fiber membranes for dialysis) and a significantly increased total surface area (over 6 m²), the pressure loss was lower than that observed in bead-packed columns. These results suggest that the favorable flow characteristics of the fiber-based adsorbent facilitate a large blood-contact area. However, further validation of the simulation results is required before fabricating an actual column. Although fibers with noncircular cross sections have been widely studied in textiles and sound-absorbing materials, ¹⁵ research on molded fibers with precisely controlled nanoporous structures at the nanoscale remains limited. In this study, we demonstrated that a noncircular fiber cross section increased the blood-contact area and improved the adsorption performance.

Furthermore, large-scale manufacturing will require verification of the reproducible cross-shaped geometry and sufficient adsorption performance of the fibers as adsorption columns.

The dense layer formed on the fiber surface was also investigated. Unlike hollow fiber membranes, which contact blood on the inner surface, the cross-shaped nanoporous fibers developed in this study contact blood on the outer surface, making the surface structure an important design factor. In the presence of a dense surface layer, the diffusion of β₂-MG into the fiber interior was hindered, reducing the adsorption performance. Notably, as shown in Figure 4, the dense layer also prevents excessive diffusion and adsorption of large-molecular-weight proteins, such as albumin. Considering molecular dimensions, β₂-MG has a long-axis length of ~4.5 nm, ¹⁶ whereas albumin measures approximately 15 nm. ¹⁷ Thus, a dense layer with an average pore diameter of ~9 nm selectively restricts albumin diffusion while allowing β₂-MG diffusion. In addition to providing selective adsorption, the dense layer contributed to the mechanical stability of the fiber owing to the high volume fraction of the supporting PMMA structure. Therefore, the thickness and structure of the dense layer must be carefully controlled according to the intended application. Irreversible albumin adsorption at the fiber surface owing to protein denaturation may cause nanopore blockage and platelet adhesion. However, previous studies have indicated that albumin adsorbed onto PMMA fibers undergoes minimal denaturation and is unlikely to promote platelet adhesion. ¹⁸ Moreover, moderate albumin removal during dialysis has been suggested to enhance hepatic albumin turnover and improve patient outcomes. ¹⁹

Adsorption rate tests using solutes of various molecular weights (Figure 3) demonstrated that cross-shaped PMMA fibers have broad adsorption capabilities, effectively adsorbing middle-molecular-weight proteins of up to ~52 kDa, with β₂-MG as the primary target. Previous in vitro studies on DRA have shown that multiple biomolecules contribute significantly to the formation and deposition of amyloid fibrils. ²⁰ Furthermore, Rosner et al. ²¹ recently identified multiple proteins ranging from 15 to 58 kDa as uremic toxins in patients with CKD, suggesting that their active removal could improve clinical outcomes.

In this study, we demonstrated that the adsorption performance of β_2-MG (molecular diameter: ~4.5 nm) was maximized when the nanopore diameter of the cross-shaped PMMA fiber was adjusted to 12–15 nm (Figure 1). In conventional dialysis, proteins such as β₂-MG are removed mainly by convection, ²² β_2-MG whereas adsorption-based systems rely primarily on diffusion; therefore, nanopore size is a critical design parameter. Specifically, smaller nanopores limit diffusion, whereas larger nanopores reduce the number of adsorption sites per unit of fiber volume. By optimizing the nanopore diameter, adsorbents and columns can be tailored to target specific proteins. Given the similarity between the nanopore diameter and molecular length of, dense adsorption within the pores is likely. Although this mechanism has been proposed previously, ²³ investigations into adsorption reversibility and irreversibility are required to fully elucidate the adsorption dynamics. Previous studies have investigated materials with precisely controlled pore sizes for artificial kidney applications, including electrospun membranes ²⁴ and nanoporous silicon membranes based on semiconductor technologies ²⁵ ; however, high manufacturing costs and limited scalability remain challenges.

Previous work 26 using PMMA fiber-packed columns included in vivo experiments with circular fibers. In a rat ECMO model, connecting a PMMA column to the bypass circuit effectively removed cytokines, reduced IL-6 levels, and suppressed systemic inflammation. Therefore, the cross-shaped PMMA fibers developed in this study, with enhanced adsorption performance, are expected to be applicable in various clinical and biomedical applications.

Conclusions

We developed cross-shaped PMMA fibers with nanopores (12–15 nm) optimized for selective β₂-microglobulin adsorption. The unique geometry increased the blood-contact area and adsorption efficiency while maintaining low pressure loss. A dense surface layer enabled selectivity by limiting large proteins, such as albumin, and permitting β₂-MG diffusion.

These fibers also adsorbed other middle-molecular-weight proteins, indicating their potential for removing uremic toxins and inflammatory mediators. Further in vitro and in vivo validation is required to confirm its clinical utility in extracorporeal circulation therapies.