Macroporous cellulose/carbon nanotube microspheres prepared by surfactant micelle swelling strategy for rapid and high-capacity adsorption of bilirubin

It remains a formidable challenge to construct high-performance haemoperfusion adsorbents with fast adsorption kinetic and high adsorption capacity for efficiently removing bilirubin from human blood. In this work, we report a facile yet efficient strategy to manufacture lysine-modified macroporous cellulose/carbon nanotube microspheres (LMCMs) by surfactant micelles swelling strategy followed by modification with lysine. The macroporous structure not only provides wide channels for fast mass transfer, but also shortens the diffusion path into meso/micropores, increasing the accessibility of mesopores and micropores. Experimental results reveal that LMCMs can remove bilirubin with fast adsorption kinetic (> 90% of its equilibrium uptake within 2 h) and high adsorption capacity of 338.14 mg/g. More importantly, the adsorbent can remove about 79% of bilirubin in rabbit serum, and the bilirubin concentration in rabbit serum decreases from 213.36 to 45.78 mg/L within 2 h, indicating a very appealing application prospect.


Introduction
Bilirubin is an endogenous compound that is released into blood due to the normal or abnormal destruction of red blood cells (Takenaka 1998).Normally, it is transported in the bloodstream to the liver for conjugation with glucuronic acid and then excreted in bile, thus the concentration of bilirubin is kept at a constant level (Chen et al. 2008;Chou and Syu 2009).However, when patients suffer from liver diseases, the generated bilirubin cannot be eliminated in time, leading to high concentration of free bilirubin in blood.These free bilirubin deposits in various tissues, and may cause mental retardation, cerebral palsy, jaundice and further lead to hepatic coma and even death (Du et al. 2017;Feng et al. 2013).
Nowadays, many techniques have been applied for the removal of excess bilirubin from blood, such as haemoperfusion, haemodialysis, and phototherapy (Guo et al. 2009).Haemoperfusion, the circulation of blood through an extracorporeal unit containing an adsorbent system, is one of the most effective techniques at present.As the heart of the circulation, many kinds of adsorbents have been designed and developed (Kavoshchian et al. 2015;Li et al. 2018;Song et al. 2019;Song et al. 2018).For example, Shi et al. (Guo et al. 2009) prepared a hollow mesoporous carbon spheres with a pore diameter of 3.8 nm by a hard template strategy, which showed an extraordinarily bilirubin adsorption with 304 mg/g, as well as high adsorption selectivity.Alexander S. Timin et al.(Timin et al. 2015) successfully developed a ureapropyl functionalized mesoporous silica adsorbents by the co-condensation of tetraethyl ortosilicate with organosilanes co-precursors, and the maximum adsorption capacity for bilirubin reaches to 0.95-2.01mg/g.Recently, our laboratory developed lysine-modi ed cellulose/carbon nanotube microspheres (LCMs) by cellulose-assisted dispersion of carbon nanotubes (CNTs) for bilirubin removal (Qiao et al. 2020b), in which cellulose serves as a base material and guarantees the blood compatibility of the composite material, and CNTs contribute to the improved mechanical strength and high adsorption capacity.The experimental results demonstrated that LCMs have high mechanical strength, excellent blood compatibility, and high bilirubin adsorption capacity of 204.12 mg/g.Although the above-mentioned works are effective, these adsorbents provide nanopores with a pore diameter of 10-100 nm for diffusing blood, so the intraparticle mass transfer exhibits a serious hindered diffusion fashion, especially for biomacromolecule such as bilirubin.The slow mass transfer means that a high dose of heparin needs to be administered by intravenous injection to prevent blood coagulation, and this may put the patients at high risk of bleeding and other severe side effects.Therefore, bilirubin adsorbents with fast mass transfer and adsorption kinetics are in strong demand.
One of the most effective methods to improve the mass transfer is to enlarge the pore size of adsorbents (Du et al. 2010;Qiao et al. 2020a).This is because macropores provide wide channels through the adsorbent for convective ow of the mobile phase, and increase the accessibility of mesopores and micropores.Several approaches have been developed for the construction of macroporous adsorbents, such as hard template method and double emulsion method.Unfortunately, the hard template methods were easily to form "island" pores, which is invalid for the mass transfer.The double emulsion method caused the formation of overlarge macropores, which would in turn lead to the adsorbents with poor mechanical property.Recently, Ma et al. proposed a surfactant reverse micelles swelling strategy to prepare macroporous microspheres for radical polymerization system, such as styrene and glycidyl methacrylate (Zhou et al. 2007a, b).However, their studies focused largely on the fabrication of hydrophobic polymer microspheres, and only a few more recent studies have reported successful preparation of hydrophilic macroporous microspheres by this method (Zhao et al. 2019).
In this work, we adopt the surfactant reverse micelles swelling strategy to prepare macroporous cellulose/CNT microspheres and followed by modi cation with lysine for bilirubin removal.The preparation process was similar to previously reported method by us (Qiao et al. 2020b), and the only difference was that a high concentration surfactant was preliminarily mixed with cellulose/CNT solution.
The experimental results demonstrate that the lysine-modi ed macroporous cellulose/CNT microspheres (LMCMs) have higher permeability, faster adsorption kinetic, and higher adsorption capacity for bilirubin compared with our previously reported LCMs without macropores.Preliminary research shows that the adsorbents have great potential for use in hemoperfusion eld.
2.2.Preparation of lysine-modi ed macroporous cellulose/CNT microspheres (LMCMs) by surfactant micelle swelling method.The preparation procedure of LMCMs is similar to that of LCMs previously reported method by us (Qiao et al. 2020b), except that a preliminarily mix of high concentration of Span-85 and cellulose/CNT solution.Brie y, 2 g of cellulose and 1 g of CNTs were dissolved in 47 g of NaOH (12 wt%)/thiourea (8 wt%)/H 2 O solution with a mass ratio of 10:8:82 solution in an ice-bath.Then, 10 g of Span-85 were drop-wise added in the cellulose /CNT dispersion and stirred for 1 h in an ice-bath.Next, 10 mL of the obtained cellulose/CNT/Span-85 mixture was poured into an oil phase consisting of 120 mL of liquid para n wax, 5g of Span-80 and 1 g of Tween-80.The resulting suspension was continued for 2h at room temperature.After that, 100 mL of 5 wt% H 2 SO 4 was poured into the suspension to induce the regeneration of macroporous cellulose/CNT microspheres.The obtained macroporous cellulose/CNT microspheres were further modi ed with lysine.The obtained lysine-modi ed macroporous cellulose/CNT microspheres were named as LMCMs.
2.3.Characterization.All samples were dried before the characterizations according to the following procedure.100 mg of wet microspheres were exchanged stepwise with 25 mL of t-BuOH solutions (20% increment).Then, the obtained microspheres were frozen in liquid nitrogen for 5 min followed by freezedried for 12 h by using a lyophilizer (FD-1A-50, Biocool).The morphology and microstructure of the samples were observed by optical microscope (PH21, Phenix) and scanning electron microscopy (Philips XL30ESEM, Eindhoven).After spraying gold on the surface of the sample, the samples were observed under 10 kV of acceleration voltage.The N 2 adsorption/desorption curves of the samples were obtained using a fully automatic speci c surface area analyzer (ASAP2020, Micromeritics).The speci c surface area and pore size distribution were determined according to the Brunauer-Emmett-Teller (BET) method and Barrett-Joyner-Halenda (BJH) method, respectively.Flow velocity dynamics experiments were conducted using the ÄKTA Explorer 100 System (Amersham Biosciences).The samples were loaded into a HR 5/10 column, and the back pressures of the sample column at different ow rates were recorded, and distilled water was used as the mobile phase.
2.4.Bilirubin adsorption experiment.For adsorption experiments, solid bilirubin was rstly dissolved in 0.2 M of NaOH solution, and then diluted with phosphatic buffer solution to a nal pH of 7.4.Subsequently, 10 mg of adsorbents was added into ten brown centrifuge tubes containing 10 mL of bilirubin solution with different concentrations (20,40,60,100,150,200,250,300,350,400,500, 600 mg/L).At intervals (0.2, 0.5, 1.0, 1.5, 2.0, 3.0, 4.0 h), the supernatants were collected to determine the bilirubin concentration at 438 nm by an ultraviolet spectrophotometer (Alpha-1900S, Puyuan).The adsorption amount was calculated with the following Eq.( 1): where q e (mg/g) is the equilibrium bilirubin adsorption capacity, c 0 and c e (mg/L) are the initial and equilibrium concentrations of bilirubin, respectively, V (L) is the volume of bilirubin solution, and m (g) is the weight of the added adsorbent.
2.5.Adsorption of bilirubin in rabbit blood.The adsorption experiment of bilirubin in rabbit blood was carried out.10 mg of adsorbents were added in 10 mL of bilirubin-enriched rabbit blood and incubated at 37℃ for 2h.Rabbit blood without adsorbent was set as a blank control to calculate the adsorption capacity of bilirubin.

Results And Discussion
It is well known that surfactant molecules at high concentration tend to aggregate and assemble to different micelles, such as such as cylindrical reverse micelle arrays, mesh phase, multilayer hexagonal and lamellar stacks (Zhou et al. 2007a;Zhou et al. 2011).The surfactant micelles have the strong ability to adsorb water or oil to form water/surfactant/oil or oil/surfactant/water emulsion (Chatjaroenporn et al. 2009).Based on the important property of surfactant, we prepared a macroporous cellulose/CNT microsphere by using high-concentration Span-85 micelles as template.The preparation process is schematically described in Fig. 1.Firstly, high concentration of Span-85 is preliminary mixed with cellulose/CNT solution to form surfactant micelles.Then, the cellulose/CNT/micelle mixture is emulsi ed to form cellulose/CNT/micelle droplets.During the emulsi cation, the hydrophobic core of the Span-85 micelles absorbs oil from external oil phase into the droplets and formed a bicontinuous oil/surfactant/water emulsion.After solidi cation, the interior oil channels are converted into macropores.Further, to improve blood compatibility and adsorption capacity, lysine as functionalized ligands are immobilized on the macroporous cellulose/CNT microspheres to obtain LMCMs (Shi et al. 2010) The external morphology and microstructure of samples were observed by optical microscope and SEM, as shown in Fig. 2. Both LCMs and LMCMs present a perfect spherical shape with a microspheres diameter of 60-90 µm.Notably, there is obvious difference in optical property between LCMs and LMCMs (Fig. 2A and C).LCMs exhibit homogeneous optical phenomenon, while LMCMs has a weak light scattering, indicating the there is a signi cant difference of the intraparticle network structure, evidenced by the following SEM (Fig. 2B and D).LCMs show a relatively smooth and compact surface structure, while HCM has macroporous structure with the pore diameter about 5 µm, con rming the effectivity of the surfactant micelles swelling strategy for constructing macropores in hydrophilic system.The unique macropoous structure can provide unimpeded mass-transfer paths for fast adsorption kinetics and increase micro/mesopore accessibility for high adsorption capacity.For the fabrication of macropores, the oil adsorption process is important to the fabrication of macroporous structure, which strongly depends on oil-adsorbing capacity of surfactant (Zhou et al. 2007b).Therefore, it can be expected that the macropore diameter of the cellulose/CNT microspheres can be tuned according to the demand in application by the control of the type or amount of surfactant.
Figure 3A shows the in uence of the ow rate of mobile phase (deionized water is used as the mobile phase) on the back pressure of the column packed LCMs and LMCMs.It can be seen that the back pressure of LMCMs was signi cantly lower than that of LCMs at the same ow rate.The low back pressure of LMCMs is due to the introduction of macroporus structure, which provides wide path for mass transfer and reduces the ow resistance.Moreover, compared with our reported macroporous chitin microspheres by solid-template method with the macropore diameter of about 0.3-1 µm, LMCMs also has lower backpressure.The low back pressure drop across the column packed LMCMs allows a high throughput operation, meaning the use of a lower dose of heparin, bene cial to avoid side effects and reduce costs.Further, based on the hydrodynamic curve, we calculated the column permeability of LMCMs and LCMs by Darcy's model (Rodrigues et al. 1995).The permeability value of LMCMs reaches to 6.22×10 − 13 m 2 (Fig. 3B), which is signi cantly higher than that of LCMs (4.02×10 − 13 m 2 ), con rming the better permeability of LMCMs.Moreover, that the back pressure of LMCMs increases linearly with the increase of ow rate, indicating LMCMs still maintain original spherical shape during operations.The result con rms that the introduction of macroporous structure has no obvious in uence on the mechanical strength of microspheres.This is the unique advantage of surfactant micelles swelling strategy method compared to the double emulsi cation method.
The porosity of LMCMs and LCMs was investigated by N 2 adsorption/desorption isotherm measurements, as shown in Fig. 4. Figure 4A shows that both materials are typical IV type adsorption/desorption isotherms with obvious H4 hysteric loops, indicating the presence of mesopores, as demonstrated by the pore size distributions.The pore size of LCMs and LMCMs mainly distributes between 10-100 nm, and pore of less than 2 nm is extremely rare (Fig. 4B and C).The Brunauer-Emmett-Teller (BET) surface area of LCMs and LMCMs reaches to 171.31 and 162.39 m 2 /g.The slight decrease in speci c surface area for LMCMs is probably due to that the surfactant micelles in uence the phased separation between cellulose/CNT and water during the solidi cation, leading to a decrease in mesopore and micropore structure (Fig. 4B and C).In spite of this, the speci c surface area of LMCMs is still comparable or higher than that of the reported cellulose-based materials (Lan et al. 2015a;Lan et al. 2015b;Lin et al. 2015).Moreover, fractal dimension (D) was calculated to analyze the porous structure according to Frenkel − Halsey − Hill (FHH) model (Sahouli et al. 1997).The D values of LCMs and LMCMs are 2.35 and 2.28 (Fig. 4D), respectively, indicating LMCMs possess more tortuous porous structure.Such irregular tortuous porous structure of MCM-NH 2 can provide more adsorption sites for high adsorption capacity.
The e ciency of LMCMs as bilirubin adsorbents for removing bilirubin has been examined by investigating its adsorption kinetics.As shown in Fig. 5, the adsorption amounts of LMCMs and LCMs increase sharply at the beginning, and then slow down gradually and nally approached the equilibrium.It can be seen that LMCMs exhibit faster adsorption rates than LCMS.The adsorption amount of LMCMs within 2 h reaches to 314.14 mg/g, far exceeds LCMs with the adsorption amount of 194.23 mg/g.Further, the experimental data were tted with pseudo-rst-order (2) and pseudo-second-order (3) models using the following equation: Where, q e and q t (mg/g) are the bilirubin adsorption amount at equilibrium of the adsorbent and the adsorption amount at time t, respectively; k 1 and k 2 (g/mg/h) are the kinetic constants of pseudo-rstorder and pseudo-second-order, respectively.All the R 2 values from pseudo-second-order (> 0.99) were higher than pseudo-rst-order, suggesting the bilirubin adsorption processes were better tted by pseudosecond-order model.The adsorption rate constant (k 2 ) of LMCMs was determined to be 0.00723 g/mg/h.This value is over 2.5 times higher than that of LCMs (0.00285 g/mg/h).Such extraordinarily fast adsorption for LMCMs is attributed to that its macroporous structure provides fast transport channels for bilirubin, which is important for e cient and safe haemoperfusion.
To investigate the bilirubin adsorption capacity of LMCMs, which is another key index for the performance criterion, the adsorption isotherms were collected with initial concentrations in the range of 20-600 mg/L.As shown in Fig. 6A.Both LMCMs and LCMs exhibit a two-stage adsorption process.With the increase of bilirubin concentration in the solution, the adsorption capacities increase rapidly at rst and then reach to equilibrium state.The equilibrium adsorption data were tted with Langmuir model ( 4), yielding a high correlation coe cient (> 0.98): q e = q m c e /(K d + c e ) (4) where, q e and q m (mg/g) are the bilirubin adsorption amount measured in the experiment and the bilirubin adsorption amount obtained by model tting, respectively.c e (mg/L) is the concentration of bilirubin in the solution after adsorption.K d (L/mg/) is the adsorption equilibrium constant of Langmuir model.
Remarkably, the maximum adsorption capacity of LMCMs reaches to 338.14 mg/g, signi cantly higher than that of LCMs (204.12 mg/g).Moreover, the speci c adsorption capacity per unit surface area of LMCMs is calculated to be 2.08 mg/m 2 , over 1.7 times than that of LCMs (1.19 mg/m 2 ) (Fig. 6B), indicating the higher surface area utilization of LMCMs.The effective surface area utilization is attributed to macroporous structure of LMCMs.The macroporous structure shortens the diffusion path into meso/micropores, increasing the accessibility of meso/micropores, thus leading to the high adsorption capacity.
To further determine the possibility of LMCMs in practical application, we used bilirubin-enriched rabbit serum rich to simulate liver failure plasma and measured its adsorption capacity, as shown in Fig. 7.After adsorption by LMCMs, the concentration of bilirubin decreases from 213.36 mg/L to 45.78 mg/L within 2h, and the concentration of bilirubin decreases by about 79%, while the control LCMs adsorb bilirubin by 67%, further con rming the superior bilirubin adsorption performance of LMCMs.The combination of high permeability, fast adsorption kinetic, and high adsorption capacity positions the LMCMs as a promising candidate for bilirubin removal.

4.
In conclusion, we have successfully developed novel lysine-macroporous cellulose/CNT microspheres (LMCMs) by surfactant micelles swelling strategy followed by modi cation with lysine.LMCMs have a macroporous structure with the pore diameter of about 5 µm and show a higher permeability of 6.22×10 − 13 m 2 compared with our previously reported LCMs without macropores.The macropores serve as a reservoir that enables rapid mass transfer to take advantage of the high surface area associated with mesopores and micropores.LMCMs exhibits superior adsorption performances for bilirubin, including fast adsorption kinetic (> 90% of its equilibrium uptake within 2h) and high adsorption capacity of 338.14 mg/g.Moreover, LMCMs can remove over 79% bilirubin within 2h from bilirubin-enriched rabbit serum, indicating the great potential of LMCMs as bilirubin adsorbents.In addition, Moreover, the advanced macroporus cellulose/CNT adsorbents can also be used as adsorbent platform to capturing other biomacromolecule, such as protein and peptide, by grafting different functional ligands, and the details of this aspect of work are currently under continuation in our laboratory.