Porous Aromatic Framework with Tailored Binding Sites and Pore Sizes as a High‐Performance Hemoperfusion Adsorbent for Bilirubin Removal

Abstract Highly efficient removal of bilirubin from blood by hemoperfusion for liver failure therapy remains a challenge in the clinical field due to the low adsorption capacity and slow adsorption kinetics of currently used bilirubin adsorbents (e.g., activated carbon and ion‐exchange resin). Recently, porous aromatic frameworks (PAFs) with high surface areas, tunable structures, and remarkable stability provide numerous possibilities to obtain satisfying adsorbents. Here, a cationic PAF with more mesopores, named iPAF‐6, is successfully constructed via a de novo synthetic strategy for bilirubin removal. The prepared iPAF‐6 exhibits a record‐high adsorption capacity of 1249 mg g−1 and can adsorb bilirubin from 150 mg L−1 to normal concentration in just 5 min. Moreover, iPAF‐6 shows a removal efficiency of 96% toward bilirubin in the presence of 50 g L−1 bovine serum albumin. It is demonstrated that positively charged aromatic frameworks and large pore size make a significant contribution to its excellent adsorption ability. More notably, iPAF‐6/polyethersulfone composite fibers or beads are fabricated for practical hemoperfusion adsorption, which also show better removal performance than commercial adsorbents. This work can offer a new possibility for designing PAF‐based bilirubin adsorbents with an appealing application prospect.


Preparation of iPAF-6/polymer composites.
To improve the practicability of iPAF-6 powders as adsorption columns, iPAF-6/polyethersulfone (PES) composite beads and fibers were fabricated. 0.75 g iPAF-6 powder was dispersed in 7.5 g DMF under the sonication for 4 h. Then, 1.75 g PES was dissolved in the solution under continuous mechanical stirring, followed by adding 1.00 g polyethylene glycol (PEG, M w = 800). After they completely dissolved, the composite solution was obtained. For iPAF-6/PES bead preparation, the composite solution was added drop-by-drop into the coagulation bath (V water /V ethanol = 1:1) to form the beads. For iPAF-6/PES fiber preparation, the composite solution was loaded into a 10 mL syringe to conduct the electrospinning process. 18 kV was applied between the cathode and anode at a distance of 15 cm with a flow rate of 1.0 mL h −1 . The iPAF-6/PES fibers were collected using a metallic rotating roller drum. The obtained beads and fibers were washed with water and methanol, and dried under vacuum at 120 °C for 12 h before use. Pure PES beads and fibers without iPAF-6 doping were also prepared with the same procedures.

Adsorption Experiments
To obtain a homogeneous bilirubin solution, a certain amount of bilirubin was initially S8 dissolved in a small volume of DMSO and 0.1 M Na 2 CO 3 solution, and then diluted with phosphate buffer saline (PBS, pH = 7.4) to get the desired concentration. In order to avoid the degradation of bilirubin, solution preparation and adsorption experiments were conducted under dark conditions. All the adsorption experiments were performed in a thermostat water bath at 37 °C.
Bilirubin adsorption kinetics. 40 mg of the adsorbents were added into 50 mL bilirubin solution (150 mg L −1 ). The mixture was shaken for 2 h at 37 °C. At appropriate time intervals, aliquots (1 mL) were taken from the mixture, and the aliquots were filtrated through a syringe filter (0.45 μm membrane filter). Then, the remaining bilirubin concentration was determined. The adsorption capacity (q) and removal efficiency (R%) of bilirubin by the adsorbent was calculated on the basis of the following equation: where C 0 and C e (mg L −1 ) are the initial and the equilibrium concentration of bilirubin in the solution, respectively. V (L) is the volume of the solution, and W (g) is the mass of the adsorbent. The kinetics data were further analyzed by the pseudo-second-order kinetic model. [2] Its linear equation is listed as follows: where q t and q e (mg g −1 ) are the adsorption capacity at time t and equilibrium time, respectively. k 2 (g mg −1 min −1 ) is the pseudo-second order model rate constant.
K d value calculation. 40 mg of the adsorbents were added into 50 mL bilirubin solution (150 mg L −1 ). After the adsorption for 2 h, the adsorbent was separated by syringe filter (0.45 μm membrane filter). The bilirubin concentrations in the resulting solutions were analyzed and the K d values were calculated as: where V is the volume of the adsorption solution (mL), m is the weight of adsorbent S9 (g), C 0 and C e the initial and equilibrium concentrations, respectively.
Bilirubin adsorption isotherms. 5 mg of the adsorbents were added into 20 mL bilirubin solutions with concentrations ranging from 35 to 500 mg L −1 . After adsorption equilibrium, the solution was filtered through a 0.45 µm syringe filter. The remaining bilirubin concentrations were analyzed and the adsorption capacities were calculated. The isotherm data were analyzed by two isotherm models, namely Langmuir and Freundlich, [3] whose linear equations are expressed as follows: Langmuir isotherm (homogeneous and monolayer adsorption): C e q e = C e q m + 1 bq m Freundlich isotherm (heterogeneous and multilayer adsorption): where q e is the equilibrium adsorption capacity (mg g −1 ), C e is the equilibrium concentration (mg L −1 ), and q m and b are Langmuir constants related to maximum adsorption capacity and binding energy, respectively; K F and n are empirical constants that indicate the Freundlich constant and heterogeneity factor, respectively.
The isotherm data were also analyzed by Zhu and Gu isotherm model (which involves two steps: the first step includes the interaction between the adsorbate molecule and the adsorbent, while the second step involves the creation of hemimicelles between adsorbed molecules through intermolecular interaction on the surface of the adsorbents) with the following equation: [4] = 1 ( 1 + 2 −1 ) where K 1 and K 2 are the equilibrium adsorption constants, and m is the average aggregation number of the hemimicelles.
The effect of albumin on bilirubin adsorption by iPAF-6. 5 mg of iPAF-6 was added into 5 mL bilirubin solution (200 mg L −1 ) with different bovine serum albumin S10 (BSA) concentration (0-50 g L −1 ). Different content of BSA was directly added to the bilirubin solution and stirred to dissolve completely before use. After adsorption equilibrium, the solution was filtered through a 0.45 µm syringe filter. The remaining bilirubin concentrations were analyzed and the removal efficiencies were calculated.
After adsorption equilibrium, the removal efficiencies of iPAF-6 towards BSA were also calculated. The BSA concentration was determined using the BCA Protein Assay Kit.
Reusability tests for iPAF-6. For the regenerative experiment, 5 mg adsorbent was where W 0 and W i are the weights of the initial samples and the samples after drying in a vacuum oven, respectively.

Computational method
To understand the intrinsic driving force for bilirubin adsorption by iPAF-5 and iPAF-6, we evaluated the interaction energies between the repeating unit as a fragment of PAFs and bilirubin using quantum chemical calculations. The 3D structures of iPAF-5, iPAF-6, and PAF-uc were built with GaussView5.0.8 and bilirubin 3D structure was download from pubchem website. All calculations were carried out using the density functional theory (DFT) with the B3LYP functiona [5,6] as implemented in the Gaussian 09 program. [7] All geometry structures were optimized with the 6-31G(d,p) basis set. On the basis of optimized geometries, the best binding structures of bilirubin with the repeating units of PAFs were searched with AutoDock Vina program, [8] in which all parameters are default. Following this, all binding structures were further optimized using B3LYP with 6-31G(d,p) basis set. To obtain more accurate energies, single-point calculations were performed based on these optimized geometries with the 6-311++G(2d,2p) basis set on all atoms. All binding energies (BE) were calculated to evaluate the adsorption strengths of bilirubin in the different PAFs. Basis set superposition error (BSSE) was considered in the calculations of binding energies.

Biocompatibility Experiments for iPAF-6
Cell viability and fluorescence images. The cells (L929 fibroblast cells or human umbilical vein endothelial cells (HUVECs)) were grown in Dulbecco's Modified S12 Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% antibiotics (penicillin-streptomycin) at 37 °C in a humidified incubator containing 5% CO 2 . A standard CCK-8 assay was used to assess the cytotoxicity of iPAF-6. Before the measurement, the iPAF-6 powders were sterilized with ethanol immersion and UV radiation. Cells at required density (1×10 5  Then, the coagulation time was measured as APTT value. For TT test, 100 μL of thrombin agent (incubated 10 min at 37 °C before use) was added to the wells, and then TT values were measured. To test PT, 100 µL Thromborel S (incubated 10 min before use) was added to the wells at 37 °C and further incubated at 37 °C for 2 min, and then the PT values were measured. Pure PPP without samples was the control group. All measurements were carried out six times.
Whole blood clotting test. The whole blood clotting test was performed according to the previous report. [9] Different amounts of iPAF-6 powders (0.5, 1. After removing the PBS, 5 mL of fresh whole blood was added. After being incubated at 37 °C for 1.5 hour, the supernatant whole blood was centrifuged for 10 minutes at 2500 rpm to obtain plasma. 40 µL of the obtained plasma was diluted for 10 times with PF4-Sample Diluent and then 200 µL of the diluted plasma was added into another Antibody Coated Well (provided by the PF4 kit). Finally, the detections were carried out according to the instruction manuals. Whole blood was used as control sample.
Complement activation test. Complement activation (C3a and C5a) was measured using enzymelinked immunosorbent assays. Fresh whole anticoagulant blood was centrifuged for 15min at 1500 rpm to obtain plasma. Then the plasma was incubated S15 with different concentrations of iPAF-6 (0.1, 0.3, 0.5, 1.0 and 2.0 mg mL −1 , respectively) for 2 h at 37 °C. A control experiment was conducted simultaneously using the same method without the addition of test samples. For the C3a test, 5 µL of the obtained plasma was diluted 500 times with C3a-sample diluent and 100 µL of the diluted plasma was then added to an antibody-coated well (provided by the C3a kit).
For the C5a test, 10 µL of the obtained plasma was diluted 10 times with C5a-sample diluent and the diluted plasma was then added into another antibody-coated well (provided by the C5a kit). All the mentioned measurements were performed according to the instruction manuals.
As shown in Figure S6, the characteristic peaks of C−Br band from bromophenyl in DBMIIB at 500~600 cm −1 and-C≡CH from TEB (or TEPB) at 3270 cm −1 almost disappear in the FTIR spectra of iPAF-5 and iPAF-6.
[10] Moreover, both spectra show the peaks of -C≡Cat 2200 cm −1 , [11] indicating the cross-coupling reactions between DBMIIB and TEB (or TEPB). In addition, the absorption peaks of -CH 2 -, -CH 3 , phenyl and imidazolium rings also exist in the spectra.               Figure S15. Bilirubin adsorption isotherms by (a) iLP and (b) PAF-uc analyzed through Zhu and Gu isotherm model.  Figure S16. EDS mapping of iPAF-6 after bilirubin adsorption from the SEM image. Ref.

S24
PDA-ordered mesoporous carbon     Previous study showed that foreign materials could induce the activation of platelets when contacted with blood, which would lead to the release of PF4 and initiate some other coagulation factors, and accelerate the formation of coagulations and thrombin. [31,32] Thus, we used PF4 concentration level to evaluate the platelet activation for iPAF-6. Figure S21. Concentrations of C3a (a) and C5a (b) in whole blood after incubation with different amounts of iPAF-6.

S27
Complement activation is considered as a trigger of the host defense mechanism, which is generated by localized inflammatory mediator. Complement activation can also reflect the blood compatibility of the materials. C3a and C5a are the activation products of complement system, which are usually used to evaluate complement activation. [33,34]