Selective Removal of the Genotoxic Compound 2-Aminopyridine in Water using Molecularly Imprinted Polymers Based on Magnetic Chitosan and β-Cyclodextrin

To develop efficient materials with enhanced adsorption and selectivity for genotoxic 2-aminopyridine in water, based on magnetic chitosan (CTs) and β-cyclodextrin (β-CD), the magnetic molecularly imprinted polymers (MMIPs) of Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP were synthesized by a molecular imprinting technique using 2-aminopyridine as a template. The selective adsorption experiments for 2-aminopyridine were performed by four analogues including pyridine, aniline, 2-amino-5-chloropyridine and phenylenediamine. Results showed the target 2-aminopyridine could be selectively adsorbed and quickly separated by the synthesized MMIPs in the presence of the above structural analogues. The coexisting ions including Na+, K+, Mg2+, Ca2+, Cl− and SO42− showed little effect on the adsorption of 2-aminopyridine. The maximum adsorption capacity of 2-aminopyridine on Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP was 39.2 mg·g−1 and 46.5 mg·g−1, respectively, which is much higher than values in previous reports. The comparison result with commercial activated carbon showed the obtained MMIPs had higher adsorption ability and selectivity for 2-aminopyridine. In addition, the synthesized MMIPs exhibited excellent performance of regeneration, which was used at least five times with little adsorption capacity loss. Therefore, the synthesized MMIPs are potential effective materials in applications for selective removal and analysis of the genotoxic compound aminopyridine from environmental water.


Introduction
Pyridine and its derivatives aminopyridines are widely used as starting materials and key intermediates in the production of pharmaceuticals, pesticides, dyes, and rubber [1]. The release of aminopyridines into the environment through various waste streams has attracted more and more attention, due to their high toxicities, carcinogenic potential and hazardous effect on ecosystem and human health [1]. Because of their relatively high solubility in water, aminopyridines can easily permeate through soil and contaminate groundwater and are difficult to degrade under aerobic and anaerobic conditions. Therefore, these substances were listed as priority pollutants by U.S. Environmental Protection Agency (U.S. EPA). Among the three monoaminopyridines, 2-aminopyridine is the most widespread in the production of various drugs, especially antihistamines and piroxicam. A previous investigation indicated that 2-aminopyridine is mutagenic, carcinogenic, and biodegradation-resistant when exposed in the environment [2]. Hence, it is necessary to The Fe 3 O 4 -chitosan particles were prepared using a one-step modifying process [28]. A typical procedure was described as follows. An aqueous solution (20 mL) containing FeCl 3 ·6H 2 O (0.7 g) and FeCl 2 ·4H 2 O (0.3 g) was dispersed ultrasonically into 100 mL of chitosan (0.5 g) solution with 0.5% (v/v) acetic acid under nitrogen atmosphere. After being stirred for 1 h at 60 • C, the solution was added to 15 mL of NH 3 ·H 2 O, drop by drop, and vigorously stirred for another 30 min. Finally, Fe 3 O 4 -chitosan magnetic particles were collected by a magnet and washed thoroughly with purified water. The products were then dried overnight at 50 • C in a vacuum oven.

Synthesis of Fe 3 O 4 -CTs@MIP
The Fe 3 O 4 -CTs@MIP was synthesized by the molecular imprinted method as described in Figure 1. Typically, 2-aminopyridine (1 mmol) was dissolved in 10 mL of acetonitrile: toluene (75: 25, v/v) in a three-necked flask. Then, MAA (4 mmol) was added and stirred for 1 h. Subsequently, 100 mL of acetonitrile: toluene (75: 25, v/v) containing 300 mg of Fe 3 O 4 -chitosan particles and 1.0 mL oleic acid were added and continuously stirred under ultrasonication for 20 min. The dispersant PVP (0.4 g), EGDMA (20 mmol) and AIBN (90 mg) were added successively to the mixture. The final mixture was purged with nitrogen while the reaction temperature increased to 60 • C and then maintained at 60 • C for 24 h. After polymerization, the Fe 3 O 4 -CTs@MIP was collected with an external magnet and Soxhlet extracted with methanol: acetic acid (9:1, v/v) until the template could not be detected in the filtrate. The obtained polymer was dried under vacuum overnight at 50 • C. Additionally, a magnetic non-imprinted polymer (Fe 3 O 4 -CTs@NIP) was synthesized with a similar procedure but without adding the template. To establish the reproducibility of the Fe 3 O 4 -CTs@MIP preparation protocols, three batches of polymer strictly following the protocol outlined above were conducted.

Synthesis of MAH-β-CD
Typically, β-CD (5.68 g) and maleic anhydride (4.9 g) were added sequentially and dissolved in 30 mL DMF. The mixture was gradually heated to 80 °C and then maintained for 10 h with vigorous stirring. After termination of the reaction, the mixture was naturally cooled to room temperature and then 30 mL of trichloromethane was slowly added. The obtained white precipitate was filtrated and rinsed several times with sufficient acetone. Finally, the product was dried under vacuum at room temperature for one day and then kept at 80 °C for 3 days.

Synthesis of MAH-β-CD Polymer Coated Magnetic Particles
Epichlorohydrin (4 mL) was dropped into sodium hydroxide solution (20 mL, 10% (w/v)) containing MAH-β-CD (2 g) and then stirred vigorously for 8 h. After the mixture solution became clear, epichlorohydrin (2 mL) was added again, and the solution was stirred overnight. The solution was concentrated and precipitated by adding cold ethanol. A fine precipitate was obtained by crushing the gummy precipitate several times with ethanol. The product was continually rinsed with ethanol and acetone and dried under strong vacuum overnight.
MAH-β-CD polymer-coated magnetic particles were prepared by modifying the method in Ref [29]. In brief, FeCl2·4H2O (0.86 g), FeCl3·6H2O (2.36 g) and MAH-β-CD polymer (1.5 g) were dispersed into 40 mL of purified water under an N2 atmosphere with vigorous stirring. Then, 5 mL of NH3·H2O was added after the solution was heated to 90 °C and maintained for 1 h. Fe3O4-MAH-β-CD magnetic particles were collected by a magnet and rinsed thoroughly with purified water. The product was then dried in a vacuum oven at 50 °C for 24 h.

Synthesis of Fe3O4-MAH-β-CD@MIP
The Fe3O4-MAH-β-CD@MIP was prepared by the molecular imprinted technique as described in Figure 2. Typically, 2-aminopyridine (0.3 mmol) and MAA (1.2 mmol) were added and dissolved in 20 mL DMF. Then, Fe3O4-MAH-β-CD particles (340 mg) were dispersed into the above solution, with ultrasonication for 20 min. EGDMA (6 mmol) and AIBN were added sequentially to the mixture under nitrogen atmosphere and reacted at 60 °C for 24 h. Finally, the product was Soxhlet

Synthesis of MAH-β-CD
Typically, β-CD (5.68 g) and maleic anhydride (4.9 g) were added sequentially and dissolved in 30 mL DMF. The mixture was gradually heated to 80 • C and then maintained for 10 h with vigorous stirring. After termination of the reaction, the mixture was naturally cooled to room temperature and then 30 mL of trichloromethane was slowly added. The obtained white precipitate was filtrated and rinsed several times with sufficient acetone. Finally, the product was dried under vacuum at room temperature for one day and then kept at 80 • C for 3 days.

Synthesis of MAH-β-CD Polymer Coated Magnetic Particles
Epichlorohydrin (4 mL) was dropped into sodium hydroxide solution (20 mL, 10% (w/v)) containing MAH-β-CD (2 g) and then stirred vigorously for 8 h. After the mixture solution became clear, epichlorohydrin (2 mL) was added again, and the solution was stirred overnight. The solution was concentrated and precipitated by adding cold ethanol. A fine precipitate was obtained by crushing the gummy precipitate several times with ethanol. The product was continually rinsed with ethanol and acetone and dried under strong vacuum overnight.
MAH-β-CD polymer-coated magnetic particles were prepared by modifying the method in Ref [29]. In brief, FeCl 2 ·4H 2 O (0.86 g), FeCl 3 ·6H 2 O (2.36 g) and MAH-β-CD polymer (1.5 g) were dispersed into 40 mL of purified water under an N 2 atmosphere with vigorous stirring. Then, 5 mL of NH 3 ·H 2 O was added after the solution was heated to 90 • C and maintained for 1 h. Fe 3 O 4 -MAH-β-CD magnetic particles were collected by a magnet and rinsed thoroughly with purified water. The product was then dried in a vacuum oven at 50 • C for 24 h.

Synthesis of Fe 3 O 4 -MAH-β-CD@MIP
The Fe 3 O 4 -MAH-β-CD@MIP was prepared by the molecular imprinted technique as described in Figure 2. Typically, 2-aminopyridine (0.3 mmol) and MAA (1.2 mmol) were added and dissolved in 20 mL DMF. Then, Fe 3 O 4 -MAH-β-CD particles (340 mg) were dispersed into the above solution, with ultrasonication for 20 min. EGDMA (6 mmol) and AIBN were added sequentially to the mixture under nitrogen atmosphere and reacted at 60 • C for 24 h. Finally, the product was Soxhlet extracted with methanol:acetic acid (9:1, v/v) and washed with purified water to ensure no more template could be detected. The obtained polymer was dried under vacuum overnight. For comparison, Fe 3 O 4 -MAH-β-CD non-imprinted polymer (Fe 3 O 4 -MAH-β-CD@NIP) without 2-aminopyridine template was also prepared for evaluation of the imprinting efficiency. In addition, three batches of Fe 3 O 4 -MAH-β-CD@MIP were conducted strictly following the protocol outlined above for the evaluation of reproducibility of preparation protocols. extracted with methanol:acetic acid (9:1, v/v) and washed with purified water to ensure no more template could be detected. The obtained polymer was dried under vacuum overnight. For comparison, Fe3O4-MAH-β-CD non-imprinted polymer (Fe3O4-MAH-β-CD@NIP) without 2-aminopyridine template was also prepared for evaluation of the imprinting efficiency. In addition, three batches of Fe3O4-MAH-β-CD@MIP were conducted strictly following the protocol outlined above for the evaluation of reproducibility of preparation protocols.

Characterization
The morphology of the MMIPs was analyzed by a scanning electron microscope (SEM) (FEI, Nova NanoSEM 450, Hillsboro, OR, USA) in secondary electron mode, with an acceleration voltage of 10 kV. The samples were coated with a thin film of gold to enable SEM imaging. The magnetic properties (saturation magnetization and coercivity) were measured using a Lakeshore 7407 vibrating sample magnetometer. X-ray powder diffraction (XRPD) patterns were collected with a PANalytical Empyrean diffractometer using monochromatic Cu Kα radiation (λ = 0.154 nm). The applied voltage and current were set to 45 kV and 40 mA, respectively. The pattern was scanned over a 2θ angle range from 20° to 70°. FT-IR spectroscopy was measured using a thermo Nicolet iS10 FT-IR spectrometer. 1 H NMR and 13 C NMR spectra were obtained from a Bruker Avance III-400M NMR spectrometer with D2O as the solvent. Thermo-gravimetric analysis (TGA) was carried out using a Q5000 thermogravimetric analyzer (TA Instruments, New Castle, DE, USA) over a temperature range of 30-700 °C. The scan rate was set to 10 °C·min −1 with dry nitrogen gas.

Adsorption Experiments
Batch adsorption experiments were conducted by adding 20 mg of adsorbents into 10 mL of 2-aminopyridine solution with initial concentrations (C0) ranging from 0.01 mg·mL −1 to 0.6 mg·mL −1 at 25 °C. After being shaken for 24 h, the adsorbents were separated with an external magnetic field,

Characterization
The morphology of the MMIPs was analyzed by a scanning electron microscope (SEM) (FEI, Nova NanoSEM 450, Hillsboro, OR, USA) in secondary electron mode, with an acceleration voltage of 10 kV. The samples were coated with a thin film of gold to enable SEM imaging. The magnetic properties (saturation magnetization and coercivity) were measured using a Lakeshore 7407 vibrating sample magnetometer. X-ray powder diffraction (XRPD) patterns were collected with a PANalytical Empyrean diffractometer using monochromatic Cu Kα radiation (λ = 0.154 nm). The applied voltage and current were set to 45 kV and 40 mA, respectively. The pattern was scanned over a 2θ angle range from 20 • to 70 • . FT-IR spectroscopy was measured using a thermo Nicolet iS10 FT-IR spectrometer. 1 H NMR and 13 C NMR spectra were obtained from a Bruker Avance III-400M NMR spectrometer with D 2 O as the solvent. Thermo-gravimetric analysis (TGA) was carried out using a Q5000 thermogravimetric analyzer (TA Instruments, New Castle, DE, USA) over a temperature range of 30-700 • C. The scan rate was set to 10 • C·min −1 with dry nitrogen gas.

Adsorption Experiments
Batch adsorption experiments were conducted by adding 20 mg of adsorbents into 10 mL of 2-aminopyridine solution with initial concentrations (C 0 ) ranging from 0.01 mg·mL −1 to 0.6 mg·mL −1 at 25 • C. After being shaken for 24 h, the adsorbents were separated with an external magnetic field, and the concentration of equilibrium solutions (C e ) were determined by UPLC. The equilibrium adsorption capacity (Q e , mg·g −1 ) was calculated: where V (mL) is the volume of the initial solution and m (g) is the weight of the adsorbent. The effect of the initial solution pH on 2-aminopyridine adsorption was investigated. The initial pH of the 2-aminopyridine solution was adjusted to the range of 2-12 by adding acid or alkaline solutions. For the kinetics study, the suspensions were shaken at 25 • C and sampled at different intervals of time in 2-aminopyridine solutions with the initial concentration of 0.1 mg·mL −1 at optimal pH value. The pseudo-first-order kinetic model, the pseudo-second-order kinetic model, and the intraparticle diffusion model were used to describe the adsorption kinetics. The adsorption thermodynamic constants of ∆H, ∆G and ∆S were also calculated.

Influence of Coexistent Ions
The adsorption performance and selective ability of MMIPs and MNIPs for 2-aminopyridine in the presence of coexisting ions such as Na + , K + , Mg 2+ , Ca 2+ , Cl − and SO 4 2− in water were investigated.

The Contrast of MMIPs and Activated Carbon in Selective Adsorption Performance
Spiked water samples (10 mL) contained 0.1 mg·mL −1 of 2-aminopyridine, and analogous using 20 mg synthesized MMIPs or granular active carbon (GAC) were applied successively. After being shaken for 4 h at 25 • C, the supernatants were centrifuged and filtered. The concentrations of free 2-aminopyridine and analogous in the filtrate were determined by UPLC.

Regeneration and Reusable Studies
Synthesized MMIPs (20 mg) were added to 10 mL of 0.1 mg·mL −1 2-aminopyridine spiked water samples. After being shaken for 4 h at 25 • C, the mixture was centrifuged and filtered. Then, the concentration of filtrate was analyzed by UPLC. The recovered MMIPs were regenerated with 5 mL of methanol/acetic acid (90:10, v/v). After drying in vacuum, it was reused in the next cycle of sorption experiments.

Removal of 2-Aminopyridine from Different Water Samples
Three types of water samples including deionized water, tap water and river water were used for evaluation of the removal performance of 2-aminopyridine. The deionized water was prepared from a Millipore system, and tap water was obtained directly from the laboratory. River water was collected from the Huangpu River in Shanghai and stored under cool and dark conditions. These water samples were filtered through a 0.22 µm filter prior to use. 2-aminopyridine was not detected in the original water samples by UPLC. MMIPs (20 mg) were suspended in 10 mL of water spiked with 0.01 mg·mL −1

Characterization of the Magnetic Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP
The surface morphologies of Fe3O4-CTs@MIP, Fe3O4-CTs@NIP, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP were characterized by SEM. Figure 4a,b showed no significant differences between Fe3O4-CTs@MIP and Fe3O4-CTs@NIP on SEM images. It can be found the morphologies with the agglomerate consisted of several spherical magnetite particles. The SEM of the prepared Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP is presented in Figure 4c,d. The Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP showed appreciable differences in morphology. The Fe3O4-MAH-β-CD@NIP crosslinking is more compact, whereas the Fe3O4-MAH-β-CD@MIP exhibited more rough and porous structures than that of Fe3O4-MAH-β-CD@NIP, indicating that the removal of template molecules resulted in possible Comparing the 1 H NMR of β-CD with MAH-β-CD, new peaks δ = 6.52, 6.20 (H 8 , H 9 ); δ = 4.57, 4.24 (H 6 ), δ = 4.03(H 5 ) were observed in MAH-β-CD. The two-dimensional (2D) HSQC 13 C-1 H plot ( Figure 3d) provides a detailed interpretation of these protons: δ H 6.52 was correlated with δ C 133.37, δ H 6.20 was correlated with δ C 126.37; therefore, δ H 6.52 and δ H 6.20 were attributed to alkene protons, δ C 133.37 and δ C 126.37 ( Figure 3c) were due to the carbon-carbon double bond of maleic anhydride unit. This result indicated that β-CD reacted with MAH. δ H 4.57 and δ H 4.24 were correlated with the same carbon δ C 64.64; therefore, the two protons belong to a methylene group, and the H-6 chemical shift change further proved that the reaction position was 6-OH. Therefore, the functionalized maleic anhydride β-CD was synthesized successfully.

Characterization of the Magnetic Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP
The surface morphologies of  The FT-IR analysis results are shown in Figure S1. In Figure S1a, a strong and wide band around 3282 cm −1 was contributed by the absorption of hydroxyl; C-O stretching vibration at 1028 cm −1 contributed to the linkage of hydroxylgroups with β-CD. The peak at 1718 cm −1 , assigned to C=O stretching vibration, and the peak at 1640 cm −1 , assigned to C=C stretching vibration, were observed in MAH-β-CD, which means that β-CD modified with vinyl carboxylic acid groups was synthesized successfully, which was consistent with previous reports [30]. Figure S1b shows the FT-IR spectra of chitosan, Fe3O4 particle, Fe3O4-CTs, Fe3O4-CTs@MIP and Fe3O4-CTs@NIP. The peak at 590 cm −1 was assigned to Fe-O stretching vibration, which proved that the Fe3O4 was successfully coated onto the polymers. For chitosan, the broad band around 3400 cm −1 was attributed to -OH and -NH2 stretching vibration. The weak band at 2920 and 2869 cm −1 was the characteristic absorbance peak of -CH2. The absorption peak at 1590 cm −1 was assigned to -NH2 bending vibration and the primary -OH alcohol bond. Additionally, the absorption peaks of symmetric stretching of C-O-C appeared at 1062 cm −1 and 1022 cm −1 . Compared with the spectra of chitosan and Fe3O4, Fe3O4-CTs exhibited a typical peak of C-O-C at 1067 cm −1 as well as a peak at 2918 cm −1 , which might be attributed to the CH2 group in chitosan. In addition, the typical strong adsorption peak at 1728 cm −1 was exhibited in Fe3O4-CTs@MIP and Fe3O4-CTs@NIP, which could be clearly attributed to the C=O stretching vibration of EGDMA and MMA. The O-H band vibration was 3432 cm −1 of Fe3O4-CTs@MIP, demonstrating the existence of a hydrogen bond in MMIP. Figure S1c shows the FT-IR spectra of the Fe3O4 particle, Fe3O4-MAH-β-CD, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP. The presence of the peak at 1723 cm −1 was attributed to C=O stretching, which confirmed that the carboxy group was incorporated into MAH-β-CD polymer. The peak at around 3400 cm −1 corresponding to the O−H stretching vibration The FT-IR analysis results are shown in Figure S1. In Figure S1a, a strong and wide band around 3282 cm −1 was contributed by the absorption of hydroxyl; C-O stretching vibration at 1028 cm −1 contributed to the linkage of hydroxylgroups with β-CD. The peak at 1718 cm −1 , assigned to C=O stretching vibration, and the peak at 1640 cm −1 , assigned to C=C stretching vibration, were observed in MAH-β-CD, which means that β-CD modified with vinyl carboxylic acid groups was synthesized successfully, which was consistent with previous reports [30]. Figure S1b shows the FT-IR spectra of chitosan, Fe 3 O 4 particle, Fe 3 O 4 -CTs, Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -CTs@NIP. The peak at 590 cm −1 was assigned to Fe-O stretching vibration, which proved that the Fe 3 O 4 was successfully coated onto the polymers. For chitosan, the broad band around 3400 cm −1 was attributed to -OH and -NH 2 stretching vibration. The weak band at 2920 and 2869 cm −1 was the characteristic absorbance peak of -CH 2 . The absorption peak at 1590 cm −1 was assigned to -NH 2 bending vibration and the primary -OH alcohol bond. Additionally, the absorption peaks of symmetric stretching of C-O-C appeared at 1062 cm −1 and 1022 cm −1 . Compared with the spectra of chitosan and Fe 3 O 4 , Fe 3 O 4 -CTs exhibited a typical peak of C-O-C at 1067 cm −1 as well as a peak at 2918 cm −1 , which might be attributed to the CH 2 group in chitosan. In addition, the typical strong adsorption peak at 1728 cm −1 was exhibited in Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -CTs@NIP, which could be clearly attributed to the C=O stretching vibration of EGDMA and MMA. The O-H band vibration was 3432 cm −1 of Fe 3 O 4 -CTs@MIP, demonstrating the existence of a hydrogen bond in MMIP. Figure S1c shows the FT-IR spectra of the Fe 3 O 4 particle, Fe 3 O 4 -MAH-β-CD, Fe 3 O 4 -MAH-β-CD@MIP and Fe 3 O 4 -MAH-β-CD@NIP. The presence of the peak at 1723 cm −1 was attributed to C=O stretching, which confirmed that the carboxy group was incorporated into MAH-β-CD polymer. The peak at around 3400 cm −1 corresponding to the O−H stretching vibration of hydroxyl groups from β-CD appeared in both two imprinted polymers. The peak at 1729 cm −1 was assigned to the stretching vibration of the carbonyl from the functional monomer MAA as well as EGDMA. The peak at 590 cm −1 , assigned to Fe-O bond vibration of Fe 3 O 4 , was observed in all polymers. Figure 5 presents the XRPD patterns of the Fe 3  of hydroxyl groups from β-CD appeared in both two imprinted polymers. The peak at 1729 cm −1 was assigned to the stretching vibration of the carbonyl from the functional monomer MAA as well as EGDMA. The peak at 590 cm −1 , assigned to Fe-O bond vibration of Fe3O4, was observed in all polymers. Figure 5 presents the XRPD patterns of the Fe3O4 particle, Fe3O4-CTs@MIP, Fe3O4-CTs@NIP, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP. Six characteristic diffraction peaks at 2θ = 30.2°, 35.5°, 43.2°, 53.5°, 57.1° and 62.8° were clearly observed in the 2θ range of 20°-70° for all the magnetic samples, which matched well with the JCPDS card No. 89-3854 for the cubic structure of Fe3O4. These characteristic diffraction peaks correspond to (220) (311) (400) (422) (511), and (440) reflection planes of the Fe3O4 crystal, respectively. The results demonstrated that the spinel structure of Fe3O4 did not undergo phase transformation during the grafting process, and Fe3O4 particles were incorporated into the prepared polymers [31]. The magnetic properties of the Fe3O4 particle, Fe3O4-MAH-β-CD@MIP, Fe3O4-MAH-β-CD@NIP, Fe3O4-CTs@MIP and Fe3O4-CTs@NIP were investigated by VSM as shown in Figure 6. It was clear that there was no hysteresis, and the symmetrical pass through the origin in the magnetization curves, indicates that all these magnetic particles retained super-paramagnetic properties. The saturation magnetization values (Ms) were 57.8 emu·g −1 , 8.2 emu·g −1 , 7.3 emu·g −1 , 4.6 emu·g −1 and 3.8 emu·g −1 . Compared with the pure Fe3O4, the saturation magnetization of polymers decreased significantly, which might be due to the decrease in the ratio of magnetic substances after the polymerization. However, Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP can be separated rapidly from suspensions under an external magnetic field. The magnetic adsorbents adhered to the walls of the vial, and the supernatant became clear and transparent as shown in the inserted photograph marked with the letter "f".  The weight percentages of Fe3O4 that adhered to Fe3O4-CTs@MIP, Fe3O4-CTs@NIP, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP were measured by TGA as shown in Figure S2. Nearly 4.0% weight loss was observed at the range of 30 °C to 700 °C for the Fe3O4 particle, which could be due to the evaporation of absorbed water as well as dehydration of the surface -OH groups. There was no significant weight loss for other polymers under 200 °C, which could be due to the stability of the polymer. The weight of these imprinted polymers decreased significantly between 300 °C and 600 °C, which might be attributed to the decomposition of carbon skeleton. Finally, there was no weight loss above 600 °C. These results demonstrated that the magnetite contents of Fe3O4-MAH-β-CD@MIPs, Fe3O4-MAH-β-CD@NIPs, Fe3O4-CTs@MIP and Fe3O4-CTs@NIP were 21.1%, 19.1%, 17.3% and 14.6%, respectively, which were consistent with the saturation magnetization value.

Effect of Solution pH on Adsorption
The effect of solution pH was investigated with the pH values ranging from 2.0 to 12.0, and the results are shown in Figure 7. It is clearly observed that the pH value had a great influence on the adsorption of 2-aminopyridine for the Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP. The compound 2-aminopyridine has two nitrogen atoms; each has lone pair electrons to donate, but the ring nitrogen atom is known to be more basic in comparison with amino nitrogen. Based on the pKa value of 2-aminopyridine (pKa = 6.86), in a low pH solution, the ring nitrogen atom of 2-aminopyridine and amino group on the surface of Fe3O4-CTs@MIP is easily protonated, resulting in the fact that the 2-aminopyridine and adsorption sites of Fe3O4-CTs@MIP carried positive charges. The strong electrostatic repulsion makes the adsorption weak. This behavior is consistent with previous reports [32,33]. In addition, the carboxyl groups grafted on Fe3O4-MAH-β-CD@MIP rarely dissociate under a low pH condition. In this case, the electrostatic interaction between polymers and 2-aminopyridine is quite weak, resulting in low adsorption ability. Therefore, they have similar weak adsorption behavior at low pH. With the increase in pH, the dissociation degree of the carboxyl groups increases, making the electrostatic repulsion weaker and electrostatic interaction stronger. Meanwhile, the carboxyl group on the surface of Fe3O4-MAH-β-CD@MIP can form hydrogen bonds with amino nitrogen of aminopyridine, which makes more binding sites available. Thus, more 2-aminopyridine could be adsorbed onto the imprinted polymers. With the help of two interactions, the adsorption capacity rises rapidly when the solution pH is in a range of 6-8. The uptake of 2-aminopyridine onto Fe3O4-CTs@MIP may mainly contribute to hydrogen bonding between the hydroxyl group in chitosan and the amino group in 2-aminopyridine. The experimental results proved that the interaction between 2-aminopyridine and Fe3O4-CTs@MIP was slightly lower than that of Fe3O4-MAH-β-CD@MIP. When pH > 8, the weakening effect of the  Figure S2. Nearly 4.0% weight loss was observed at the range of 30 • C to 700 • C for the Fe 3 O 4 particle, which could be due to the evaporation of absorbed water as well as dehydration of the surface -OH groups. There was no significant weight loss for other polymers under 200 • C, which could be due to the stability of the polymer. The weight of these imprinted polymers decreased significantly between 300 • C and 600 • C, which might be attributed to the decomposition of carbon skeleton. Finally, there was no weight loss above 600 • C. These results demonstrated that the magnetite contents of Fe 3 O 4 -MAH-β-CD@MIPs, Fe 3 O 4 -MAH-β-CD@NIPs, Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -CTs@NIP were 21.1%, 19.1%, 17.3% and 14.6%, respectively, which were consistent with the saturation magnetization value.

Effect of Solution pH on Adsorption
The effect of solution pH was investigated with the pH values ranging from 2.0 to 12.0, and the results are shown in Figure 7. It is clearly observed that the pH value had a great influence on the adsorption of 2-aminopyridine for the Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP. The compound 2-aminopyridine has two nitrogen atoms; each has lone pair electrons to donate, but the ring nitrogen atom is known to be more basic in comparison with amino nitrogen. Based on the pKa value of 2-aminopyridine (pKa = 6.86), in a low pH solution, the ring nitrogen atom of 2-aminopyridine and amino group on the surface of Fe 3 O 4 -CTs@MIP is easily protonated, resulting in the fact that the 2-aminopyridine and adsorption sites of Fe 3 O 4 -CTs@MIP carried positive charges. The strong electrostatic repulsion makes the adsorption weak. This behavior is consistent with previous reports [32,33]. In addition, the carboxyl groups grafted on Fe 3 O 4 -MAH-β-CD@MIP rarely dissociate under a low pH condition. In this case, the electrostatic interaction between polymers and 2-aminopyridine is quite weak, resulting in low adsorption ability. Therefore, they have similar weak adsorption behavior at low pH. With the increase in pH, the dissociation degree of the carboxyl groups increases, making the electrostatic repulsion weaker and electrostatic interaction stronger. Meanwhile, the carboxyl group on the surface of Fe 3 O 4 -MAH-β-CD@MIP can form hydrogen bonds with amino nitrogen of aminopyridine, which makes more binding sites available. Thus, more 2-aminopyridine could be adsorbed onto the imprinted polymers. With the help of two interactions, the adsorption capacity rises rapidly when the solution pH is in a range of 6-8. The uptake of 2-aminopyridine onto Fe 3 O 4 -CTs@MIP may mainly contribute to hydrogen bonding between the hydroxyl group in chitosan and the amino group in 2-aminopyridine. The experimental results proved that the interaction between 2-aminopyridine and Fe 3 O 4 -CTs@MIP was slightly lower than that of Fe 3 O 4 -MAH-β-CD@MIP. When pH > 8, the weakening effect of the electrostatic interaction between polymers and 2-aminopyridine was displayed obviously, leading to a lower adsorption capacity. Therefore, the optimal pH value of 8 was chosen for subsequent adsorption studies. electrostatic interaction between polymers and 2-aminopyridine was displayed obviously, leading to a lower adsorption capacity. Therefore, the optimal pH value of 8 was chosen for subsequent adsorption studies.

Adsorption Isotherm
The Langmuir [34] and Freundlich isotherm models [35], which can be expressed as Equations (2) and (3), respectively, are widely adopted for adsorption equilibrium data analysis.
e L e m L e C K 1 where Qe (mg·g −1 ) is the equilibrium adsorption capacity, Ce (mg·mL −1 ) is the equilibrium concentration of 2-aminopyridine in aqueous solution, Qm (mg·g −1 ) is the maximum adsorption capacity. KL represents the Langmuir constant, KF is the Freundlich constant and 1/n is the heterogeneity factor.
The adsorption amounts of 2-aminopyridine versus the corresponding equilibrium concentration at 298 K were plotted as adsorption isotherms in Figure 8. With increasing 2-aminopyridine concentration, the adsorption capacity of 2-aminopyridine was increased clearly. The calculated isotherm parameters accompanied with standard errors of Fe3O4-CTs@MIP, Fe3O4-CTs@NIP, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP for 2-aminopyridine adsorption was summarized in Table 1. Standard errors of three experiment replicates are within the range from 2.8 to 6.4%. By comparing the regression coefficients (R 2 ), the Langmuir model fitted the data better compared to the Freundlich model. This indicated that adsorption might occur on the homogeneity active sites of the absorbent surface. The maximum adsorption capacities of 2-aminopyridine binding to Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP calculated by the Langmuir isotherm model were 39.2 mg·g −1 and 46.5 mg·g −1 , respectively, which were higher than those on the Fe3O4-CTs@NIP and Fe3O4-MAH-β-CD@NIP. This revealed that the synthesized MMIP had more specific sites than the MNIP for the template molecules.

Adsorption Isotherm
The Langmuir [34] and Freundlich isotherm models [35], which can be expressed as Equations (2) and (3), respectively, are widely adopted for adsorption equilibrium data analysis.
where Q e (mg·g −1 ) is the equilibrium adsorption capacity, C e (mg·mL −1 ) is the equilibrium concentration of 2-aminopyridine in aqueous solution, Q m (mg·g −1 ) is the maximum adsorption capacity. K L represents the Langmuir constant, K F is the Freundlich constant and 1/n is the heterogeneity factor. The adsorption amounts of 2-aminopyridine versus the corresponding equilibrium concentration at 298 K were plotted as adsorption isotherms in Figure 8. With increasing 2-aminopyridine concentration, the adsorption capacity of 2-aminopyridine was increased clearly. The calculated isotherm parameters accompanied with standard errors of Fe 3 O 4 -CTs@MIP, Fe 3 O 4 -CTs@NIP, Fe 3 O 4 -MAH-β-CD@MIP and Fe 3 O 4 -MAH-β-CD@NIP for 2-aminopyridine adsorption was summarized in Table 1. Standard errors of three experiment replicates are within the range from 2.8 to 6.4%. By comparing the regression coefficients (R 2 ), the Langmuir model fitted the data better compared to the Freundlich model. This indicated that adsorption might occur on the homogeneity active sites of the absorbent surface. The maximum adsorption capacities of 2-aminopyridine binding to Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP calculated by the Langmuir isotherm model were 39.2 mg·g −1 and 46.5 mg·g −1 , respectively, which were higher than those on the Fe 3 O 4 -CTs@NIP and Fe 3 O 4 -MAH-β-CD@NIP. This revealed that the synthesized MMIP had more specific sites than the MNIP for the template molecules. The adsorption amount of 2-aminopyridine on Fe3O4-MAH-β-CD@MIP was higher than Fe3O4-CTs@MIP, which could be due to more hydrogen bonding and electrostatic interaction between Fe3O4-MAH-β-CD@MIP and 2-aminopyridine molecules. Significantly, the adsorption capacities of all the synthesized materials in this work were much higher than those of the previously reported conventional imprinted polymers [16,27]. However, the MIH-FRP hydrogel with high adsorption capacity of 3-aminopyridine was prepared in aqueous media, bringing a stronger electrostatic interaction between the template and monomer self-assembly. this provides a new approach for the preparation of high performance imprinted hydrogel [36]. In our work, MMIPs were prepared in organic solvents through imprinting technology based on novel supporting materials. They have promising imprinting inside the MMIPs network that has strong hydrogen bonding and electrostatic interaction between 2-aminopyridine and polymers. Moreover, the MMIPs are magnetic and can be separated rapidly from suspensions under an external magnetic field. The comparative results are shown in Table 2. To evaluate the reproducibility of MMIP preparation, the adsorption properties of three batches of MMIP were investigated. Table 3 shows the batch reproducibility data of Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP. The results reveal that Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP for 2-aminopyridine could be prepared consistently.  The adsorption amount of 2-aminopyridine on Fe 3 O 4 -MAH-β-CD@MIP was higher than Fe 3 O 4 -CTs@MIP, which could be due to more hydrogen bonding and electrostatic interaction between Fe 3 O 4 -MAH-β-CD@MIP and 2-aminopyridine molecules. Significantly, the adsorption capacities of all the synthesized materials in this work were much higher than those of the previously reported conventional imprinted polymers [16,27]. However, the MIH-FRP hydrogel with high adsorption capacity of 3-aminopyridine was prepared in aqueous media, bringing a stronger electrostatic interaction between the template and monomer self-assembly. this provides a new approach for the preparation of high performance imprinted hydrogel [36]. In our work, MMIPs were prepared in organic solvents through imprinting technology based on novel supporting materials. They have promising imprinting inside the MMIPs network that has strong hydrogen bonding and electrostatic interaction between 2-aminopyridine and polymers. Moreover, the MMIPs are magnetic and can be separated rapidly from suspensions under an external magnetic field. The comparative results are shown in Table 2. To evaluate the reproducibility of MMIP preparation, the adsorption properties of three batches of MMIP were investigated. Table 3 Figure S3. Table 4. Clearly, one finds that the adsorption of 2-aminopyridine decreased as the temperature increased. The Langmuir equilibrium constant (K L ) decreased with increasing temperature, indicating that the binding ability of 2-aminopyridine on Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP was higher at lower temperature. Furthermore, the calculated Langmuir maximum adsorption capacities of Fe 3 O 4 -MAH-β-CD@MIP for 2-aminopyridine were higher than those of Fe 3 O 4 -CTs@MIP regardless of the experimental temperature.  Figure 9 shows a time-dependent adsorption dynamics of 2-aminopyridine on Fe 3 O 4 -CTs@MIP, Fe 3 O 4 -MAH-β-CD@MIP and MNIPs at optimized pH and room temperature. The adsorption of all these four adsorbents was fast in the first 20 min and then achieved equilibrium after approximately 30 min. This was mainly due to the existence of a large number of active sites on the surface of adsorbents at the beginning. However, with the increase in the contact time, the slope of the adsorption capacity increased slowly, which might be attributed to the gradual occupation of the binding sites on the adsorbent surface by aminopyridine, resulting in decreased adsorption rate at the later stage. In addition, it was observed that the capacity of the adsorbed 2-aminopyridine on Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP at any time was considerably greater than that of the corresponding MNIPs. This was mainly due to the specific recognition effect and affinity binding sites in imprinted polymers. The pseudo-first order [37] and pseudo-second order models [38] were used to evaluate the adsorption kinetics, which are described in the following Equations (4) and (5), respectively:

Adsorption Kinetics
where k 1 (min −1 ) and k 2 (g·mg −1 ·min −1 ) are the rate constant of pseudo-first order and pseudo-second order adsorption, respectively; Q e (mg·g −1 ) and Q t (mg·g −1 ) are the adsorption capacity at equilibrium and time t, respectively. corresponding MNIPs. This was mainly due to the specific recognition effect and affinity binding sites in imprinted polymers. The pseudo-first order [37] and pseudo-second order models [38] were used to evaluate the adsorption kinetics, which are described in the following Equations (4) and (5), respectively: where k1 (min −1 ) and k2 (g·mg −1 ·min −1 ) are the rate constant of pseudo-first order and pseudo-second order adsorption, respectively; Qe (mg·g −1 ) and Qt (mg·g −1 ) are the adsorption capacity at equilibrium and time t, respectively. The parameters of adsorption kinetics are summarized in Table S1. The regression curves of the pseudo-first-order and pseudo-second-order rate equation of linear forms for 2-aminopyridine adsorption are shown in Figures S4 and S5. It is known that the pseudo-second order kinetic model is assumed for the chemical reaction mechanisms, and that the adsorption rate is controlled by chemical adsorption between chemical binding sites on the surface of adsorbent and adsorbate. By comparison of the data, the adsorption processes of 2-aminopyridine on adsorbents were better fitted by the pseudo-second order model with a good correlation coefficient (R 2 > 0.99). Therefore, the adsorption behavior of 2-aminopyridine onto adsorbents belonged to the pseudo-second order model, and the adsorption process was mainly a chemical process. The adsorption rate constant k2 for 2-aminopyridine on Fe3O4-MAH-β-CD@MIP was greater than that on Fe3O4-CTs@MIP, indicating a higher rate for 2-aminopyridine removal by Fe3O4-MAH-β-CD@MIP. Furthermore, compared with the pseudo-first order model, the calculated Q values (Qe, cal) obtained by the pseudo-second order model were highly consistent with the experimental data. The intraparticle diffusion model was proposed for the interpretation of adsorption progress of aminopyridine intraparticle of the imprinted polymers. Equation (6) is given by Weber and Morris [39] as below: The parameters of adsorption kinetics are summarized in Table S1. The regression curves of the pseudo-first-order and pseudo-second-order rate equation of linear forms for 2-aminopyridine adsorption are shown in Figures S4 and S5. It is known that the pseudo-second order kinetic model is assumed for the chemical reaction mechanisms, and that the adsorption rate is controlled by chemical adsorption between chemical binding sites on the surface of adsorbent and adsorbate. By comparison of the data, the adsorption processes of 2-aminopyridine on adsorbents were better fitted by the pseudo-second order model with a good correlation coefficient (R 2 > 0.99). Therefore, the adsorption behavior of 2-aminopyridine onto adsorbents belonged to the pseudo-second order model, and the adsorption process was mainly a chemical process. The adsorption rate constant k 2 for 2-aminopyridine on Fe 3 O 4 -MAH-β-CD@MIP was greater than that on Fe 3 O 4 -CTs@MIP, indicating a higher rate for 2-aminopyridine removal by Fe 3 O 4 -MAH-β-CD@MIP. Furthermore, compared with the pseudo-first order model, the calculated Q values (Q e , cal) obtained by the pseudo-second order model were highly consistent with the experimental data. The intraparticle diffusion model was proposed for the interpretation of adsorption progress of aminopyridine intraparticle of the imprinted polymers. Equation (6) is given by Weber and Morris [39] as below: Q t = k i t 1/2 + C where C (mg·g −1 ) is a constant which reflects the thickness of the boundary layer and k i is the intra-particle diffusion rate constant (mg·g −1 ·min −1/2 ). If the curve of Q t against t 1/2 passes through the origin, the adsorption process is only controlled by intra-particle diffusion. Otherwise, some other mechanisms may be involved in the adsorption process. Curves of Q t against t 1/2 for 2-aminopyridine adsorption onto absorbents are presented in Figure 10. It exhibited three linear plots for 2-aminopyridine adsorption, indicating three steps influenced in the adsorption process. In the initial fast step, the diffusion of 2-aminopyridine may cause the rapid adsorption. Then, intraparticle diffusion into mesopores and micropores in the gradual adsorption stage is the rate determining step. In the final stage, the slowdown of intraparticle diffusion may be attributed to the relatively low content of residual 2-aminopyridine in the solution [40,41]. Therefore, the intraparticle diffusion was involved in the adsorption process but was not the only rate determining step. where C (mg·g −1 ) is a constant which reflects the thickness of the boundary layer and ki is the intra-particle diffusion rate constant (mg·g −1 ·min −1/2 ). If the curve of Qt against t 1/2 passes through the origin, the adsorption process is only controlled by intra-particle diffusion. Otherwise, some other mechanisms may be involved in the adsorption process. Curves of Qt against t 1/2 for 2-aminopyridine adsorption onto absorbents are presented in Figure 10. It exhibited three linear plots for 2-aminopyridine adsorption, indicating three steps influenced in the adsorption process. In the initial fast step, the diffusion of 2-aminopyridine may cause the rapid adsorption. Then, intraparticle diffusion into mesopores and micropores in the gradual adsorption stage is the rate determining step. In the final stage, the slowdown of intraparticle diffusion may be attributed to the relatively low content of residual 2-aminopyridine in the solution [40,41]. Therefore, the intraparticle diffusion was involved in the adsorption process but was not the only rate determining step.  Figure S6 shows that, as the temperature increased, the values of ln(Ca/Ce) decreased, illustrating the exothermic nature of the 2-aminopyridine adsorption process. The values of ln(Ca/Ce) at various temperatures were calculated using Equation (7) as proposed by Van't Hoff [42]:

Adsorption Thermodynamics
The Gibbs free energy ΔG (kJ·mol −1 ) and entropy change ΔS of adsorption are described by Equations (8) and (9), respectively, as follows [43]:  Figure S6 shows that, as the temperature increased, the values of ln(C a /C e ) decreased, illustrating the exothermic nature of the 2-aminopyridine adsorption process. The values of ln(C a /C e ) at various temperatures were calculated using Equation (7) as proposed by Van't Hoff [42]:

Adsorption Thermodynamics
The Gibbs free energy ∆G (kJ·mol −1 ) and entropy change ∆S of adsorption are described by Equations (8) and (9), respectively, as follows [43]: where, C a and C e are the equilibrium concentrations of 2-aminopyridine on the adsorbents and solution at different temperatures, respectively. R is the universal gas constant (8.314 J/mol·K) and T is the absolute temperature (K). The enthalpy change (∆H, kJ·mol −1 ) was calculated from the slope of the line plotted by ln(C a /C e ) against 1/T. The calculated thermodynamic parameters are presented in Table 5. It was clearly observed that the values of ∆G were negative for both the adsorption process on Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP at various temperatures, illustrating that the adsorption process on the two adsorbents towards 2-aminopyridine were spontaneous. In addition, the observed negative values of ∆G decreased as the temperature increased, indicating that the higher temperature has an adverse effect on the adsorption. The negative values of entropy ∆S suggested the decreased randomness at the solid-liquid interface during the adsorption of 2-aminopyridine onto both absorbents. Finally, the obtained negative ∆H values also further confirmed the exothermic nature of the adsorption process [44].   Figure 11 shows the adsorption selectivity of MMIP and MNIP for 2-aminopyridine and its analogues. The results demonstrated that all the MMIPs have higher selective affinity for 2-aminopyridine and its analogues compared with that of MNIPs except for pyridine. The differences between MMIPs and MNIPs are due to the recognition effect of MMIPs. The exception for pyridine can be explained by its structure: there was no substitute group on the ring. The result also showed that both Fe 3 O 4 -MAH-β-CD@MIP and Fe 3 O 4 -CTs@MIP exhibited much higher selectivity for 2-aminopyridine than those of the other four analogs. The reason can be attributed to the effect of imprinting with 2-aminopyridine, which is a template of monosubstituted pyridine. It is clearly suggests the presence of specific imprinting sites that favored the adsorption of the target molecule. For di-substituted pyridine 2-A-5-CP and aminobenzene compounds of aniline and phenylenediamine, the relative lower adsorption amounts may also be related to the steric hindrance of the two substitute groups and the ring structure without nitrogen. Therefore, it can be indicated that Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP have good selectivity and specific recognition toward 2-aminopyridine due to the imprinting effect, even with the interference of competitive analogues.

Influence of Coexistent Ions
In the presence of coexistent ions including Na + , K + , Mg 2+ , Ca 2+ , Cl − and SO4 2− in water samples, the binding efficiency and selective ability of Fe3O4-CTs@MIP, Fe3O4-CTs@NIP, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP for 2-aminopyridine were investigated and the results are presented in Figure 12. It is observed that the binding capacity of synthesized MMIPs and MNIPs shows no significant change in both the absence and presence of various ions in water samples. This means that the binding was hardly influenced by the coexisting ions in water. Thus, the MMIPs synthesized in our work were possibly involved in detecting or removing 2-aminopyridine in water.

Influence of Coexistent Ions
In the presence of coexistent ions including Na + , K + , Mg 2+ , Ca 2+ , Cl − and SO 4

Influence of Coexistent Ions
In the presence of coexistent ions including Na + , K + , Mg 2+ , Ca 2+ , Cl − and SO4 2− in water samples, the binding efficiency and selective ability of Fe3O4-CTs@MIP, Fe3O4-CTs@NIP, Fe3O4-MAH-β-CD@MIP and Fe3O4-MAH-β-CD@NIP for 2-aminopyridine were investigated and the results are presented in Figure 12. It is observed that the binding capacity of synthesized MMIPs and MNIPs shows no significant change in both the absence and presence of various ions in water samples. This means that the binding was hardly influenced by the coexisting ions in water. Thus, the MMIPs synthesized in our work were possibly involved in detecting or removing 2-aminopyridine in water.

Comparison of MIPs and Activated Carbon in Selective Adsorption Performance
For comparison, the adsorption and selective performance of the synthesized MMIPs were compared with those of commercial granular activated carbon (GAC) for the treatment of 2-aminopyridine and analogues in water. GAC is activated carbon with particle sizes predominantly greater than 80 mesh and a surface area between 900 and 1110 m 2 ·g −1 . As shown in Figure 13, owing to the imprinting effect, two MMIPs showed a higher recognition and selective ability to 2-aminopyridine compared with activated carbon. The Fe 3 O 4 -MAH-β-CD@MIP exhibited the greatest affinity among the three adsorbents. The commercial activated carbon material showed a similar adsorption ability to all three organic substances, and the selectivity was not obvious.

Comparison of MIPs and Activated Carbon in Selective Adsorption Performance
For comparison, the adsorption and selective performance of the synthesized MMIPs were compared with those of commercial granular activated carbon (GAC) for the treatment of 2-aminopyridine and analogues in water. GAC is activated carbon with particle sizes predominantly greater than 80 mesh and a surface area between 900 and 1110 m 2 ·g −1 . As shown in Figure 13, owing to the imprinting effect, two MMIPs showed a higher recognition and selective ability to 2-aminopyridine compared with activated carbon. The Fe3O4-MAH-β-CD@MIP exhibited the greatest affinity among the three adsorbents. The commercial activated carbon material showed a similar adsorption ability to all three organic substances, and the selectivity was not obvious.

Regeneration and Reusable Studies
The regeneration properties of Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP after adsorption were investigated to understand the possibility of further practical application. As shown in Figure 14, the synthesized MMIPs can be regenerated after washing with methanol/acetic acid (90:10, v/v). Their adsorption efficiencies were stable for up to five adsorption-regeneration cycles without a considerable decrease for 2-aminopyridine. After the fifth regeneration, the reduction in the adsorption capacity of Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP for 2-aminopyridine was only 6.9% and 7.6%, respectively. The result indicated that Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP are desirable for potential application in future.

Regeneration and Reusable Studies
The regeneration properties of Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP after adsorption were investigated to understand the possibility of further practical application. As shown in Figure 14, the synthesized MMIPs can be regenerated after washing with methanol/acetic acid (90:10, v/v). Their adsorption efficiencies were stable for up to five adsorption-regeneration cycles without a considerable decrease for 2-aminopyridine. After the fifth regeneration, the reduction in the adsorption capacity of Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP for 2-aminopyridine was only 6.9% and 7.6%, respectively. The result indicated that Fe 3 O 4 -CTs@MIP and Fe 3 O 4 -MAH-β-CD@MIP are desirable for potential application in future.

Removal of 2-Aminopyridine with MMIPs in Different Water Samples
For evaluating the possibility of practical applications, the adsorption performance of the synthesized MMIPs for the genotoxic compound 2-aminopyridine in different water samples was investigated. Usually, for the discharged water after routine treatment, the level of the pollutant may be about 1-10 mg·L −1 . Therefore, the experimental water samples of spiked 10 mg·L −1 2-aminopyridine in deionized water, tap water and river water samples were used for evaluation.
As shown in Figure 15, there was no significant difference for the adsorption amount of synthesized MMIP for 2-aminopyridine in deionized water, tap water and real river water samples. The results indicated that, owing to the existence of specific binding sites in Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP, the MMIPs have a good adsorption performance for 2-aminopyridine with high selectivity and efficiency in all kinds of experimental water samples.

Conclusions
In this study, two novel magnetic molecular imprinted polymers based on chitosan and MAH-β-CD were successfully synthesized by a molecular imprinting technique for the selective removal and magnetic separation of 2-aminopyridine in aqueous solutions. Adsorption

Removal of 2-Aminopyridine with MMIPs in Different Water Samples
For evaluating the possibility of practical applications, the adsorption performance of the synthesized MMIPs for the genotoxic compound 2-aminopyridine in different water samples was investigated. Usually, for the discharged water after routine treatment, the level of the pollutant may be about 1-10 mg·L −1 . Therefore, the experimental water samples of spiked 10 mg·L −1 2-aminopyridine in deionized water, tap water and river water samples were used for evaluation.
As shown in Figure 15, there was no significant difference for the adsorption amount of synthesized MMIP for 2-aminopyridine in deionized water, tap water and real river water samples. The results indicated that, owing to the existence of specific binding sites in Fe 3 O 4 -CTs@MIP and

Removal of 2-Aminopyridine with MMIPs in Different Water Samples
For evaluating the possibility of practical applications, the adsorption performance of the synthesized MMIPs for the genotoxic compound 2-aminopyridine in different water samples was investigated. Usually, for the discharged water after routine treatment, the level of the pollutant may be about 1-10 mg·L −1 . Therefore, the experimental water samples of spiked 10 mg·L −1 2-aminopyridine in deionized water, tap water and river water samples were used for evaluation.
As shown in Figure 15, there was no significant difference for the adsorption amount of synthesized MMIP for 2-aminopyridine in deionized water, tap water and real river water samples. The results indicated that, owing to the existence of specific binding sites in Fe3O4-CTs@MIP and Fe3O4-MAH-β-CD@MIP, the MMIPs have a good adsorption performance for 2-aminopyridine with high selectivity and efficiency in all kinds of experimental water samples.

Conclusions
In this study, two novel magnetic molecular imprinted polymers based on chitosan and MAH-β-CD were successfully synthesized by a molecular imprinting technique for the selective removal and magnetic separation of 2-aminopyridine in aqueous solutions. Adsorption

Conclusions
In this study, two novel magnetic molecular imprinted polymers based on chitosan and MAH-β-CD were successfully synthesized by a molecular imprinting technique for the selective removal and magnetic separation of 2-aminopyridine in aqueous solutions. Adsorption experimental results showed that the synthesized MMIPs materials had high adsorption capacity and good selectivity for 2-aminopyridine owing to the imprinting effect. Compared with other sorbents reported in the literature, the adsorption performance of MIPs prepared in this work showed a significant improvement. The adsorption capacity of Fe 3 O 4 -MAH-β-CD@MIP was higher than that of Fe 3 O 4 -CTs@MIP, which could be attributed to the stronger action on hydrogen bonding and electrostatic attraction between Fe 3 O 4 -MAH-β-CD@MIP and the 2-aminopyridine target molecule. In addition, the synthesized MMIPs exhibited higher specific recognition and selectivity to 2-aminopyridine in the presence of interference substances and showed excellent performance of regeneration. It can be used at least five times with little adsorption capacity loss. Therefore, the imprinted polymers are expected to have promising application for the removal and analysis of 2-aminopyridine in environmental water. However, in order to have a larger scale and practical application in future, numerous different factors should be considered, and more work is required.