Fabrication of Hypericin Imprinted Polymer Nanospheres via Thiol-Yne Click Reaction

To fabricate molecularly imprinted polymer nanospheres via click reaction, five different clickable compounds were synthesized and two types of click reactions (azide-alkyne and thiol-yne) were explored. It was found that molecularly imprinted polymer nanospheres could be successfully synthesized via thiol-yne click reaction using 3,5-diethynyl-pyridine (1) as the monomer, tris(3-mercaptopropionate) (tri-thiol, 5) as the crosslinker, and hypericin as the template (MIP–NSHs). The click polymerization completed in merely 4 h to produce the desired MIP–NSHs, which were characterized by FTIR, SEM, DLS, and BET, respectively. The reaction conditions for adsorption capacity and selectivity towards hypericin were optimized, and the MIP–NSHs synthesized under the optimized conditions showed a high adsorption capacity (Q = 6.03 μmol·g−1) towards hypericin. The imprinting factors of MIP–NSHs towards hypericin, protohypericin, and emodin were 2.44, 2.88, and 2.10, respectively.


Introduction
As an important type of artificial receptor, molecularly imprinted polymers (MIPs) [1,2], proposed in the latter part of the last century, have become increasingly attractive to both the academic community and in industry due to their high affinity towards given molecules (templates), high chemical stability, and low cost [3], and have been applied widely in chromatography [4], solid phase extraction (SPE) [5], separation [6], immunoassay [7], sensors [8], catalysis [9], and drug delivery [10][11][12]. MIPs are synthesized via the copolymerization of functional monomers and crosslinkers in the presence of template molecules that form assemblies with functional monomers driven by covalent or non-covalent interactions. Sequential removal of the template molecules from the crosslinked polymer networks leaves the recognition cavities complementary to the shape, size, and position of the functional groups, which then show specific recognition and rebinding capacity towards the template molecules upon re-exposure [13].
The most common physical form of MIPs is polymer monolith obtained via bulk polymerization, which usually produces irregular polymeric particles in a range of 5-100 µm in low yield (less than 50%) [13,14] through a tedious process of crushing, grinding and sieving. Although the method is simple, it causes considerable wastage of the MIPs. More importantly, the irregularities of the resultant particles, in terms of size, shape, and architecture, produced by this method make sample handling difficult and decrease separation efficiency, which limit the application of MIPs, particularly in SPE or chromatography [15][16][17]. In contrast, spherical particles with a well-defined structure and monodispersity are more desirable due to their fast mass transfer rate, better separation performance methods have been successfully established for the fabrication of spherical MIPs beads, including inverse suspension polymerization [20], emulsion polymerization [21], solution polymerization [22], multi-step swelling polymerization [23], and precipitation polymerization [13,24]. However, these methods are either complicated in manipulation, require a long time to complete polymerization, or require surfactants (which are difficult to remove from MIPs) to prevent aggregation of polymer particles formed in the process of polymerization. Development of new methods that can produce MIP nanospheres efficiently with well-defined architecture in a high yield is therefore still much desired. Click reaction, introduced by Sharpless in 2001 [25], has been recognized as a popular and powerful tool in controlling macromolecular architecture due to its mild, fast, highly efficient and specific reaction characteristics [26]. However, click reaction was not applied in the field of molecular imprinting technology until 2011, when Ye and his co-workers [27] prepared clickable molecularly imprinted core-shell nanoparticles using a simple one-pot precipitation polymerization with sequential addition of Scheme 1. (a) Illustration of hydrogen bonding of template, hypericin with functional monomer 1; (b) Schematic illustration of fabricating MIP-NSHs with 1 as monomer and 5 as crosslinker.

Preparation of Imprinted Polymer Nanospheres towards Hypericin and Non-Imprinted Polymer Nanospheres
Typically, polymer nanospheres imprinted towards hypericin (MIP-NSHs) were prepared as follows. To a quartz tube, 0.1 mmol functional monomer, 0.1 mmol cross-linker, 0.031 mmol DMPA, 2 mL acetone were added, and the mixture was subjected to ultrasonication for 10 min, followed by adding 5 μmol template. The click polymerization was performed for 4 h under UV light at a wavelength of 350 nm at ambient temperature under N2 and stirring. After the reaction, the resulting beads were collected, filtered, and extracted to remove hypericin. Then the polymer beads were allowed to dry naturally and grounded with a pestle for 15 min to obtain MIP-NSHs.
The extraction procedure: The nanospheres were extracted with a soxhlet extractor using 10% acetic acid in acetone for 48 h, thereafter the nanoshperes were ultrasonicated in 5 mL 20% acetic acid in acetone for 20 min and then centrifuged; the resulting supernatant was monitiored using HPLC. The ultrasonication step was repeated until no hypericin could be detected in the supernatant.
Control polymer nanospheres (NIP-NSs) were prepared under identical conditions but in the absence of hypericin.

Preparation of Imprinted Polymer Nanospheres towards Hypericin and Non-Imprinted Polymer Nanospheres
Typically, polymer nanospheres imprinted towards hypericin (MIP-NSHs) were prepared as follows. To a quartz tube, 0.1 mmol functional monomer, 0.1 mmol cross-linker, 0.031 mmol DMPA, 2 mL acetone were added, and the mixture was subjected to ultrasonication for 10 min, followed by adding 5 µmol template. The click polymerization was performed for 4 h under UV light at a wavelength of 350 nm at ambient temperature under N 2 and stirring. After the reaction, the resulting beads were collected, filtered, and extracted to remove hypericin. Then the polymer beads were allowed to dry naturally and grounded with a pestle for 15 min to obtain MIP-NSHs.
The extraction procedure: The nanospheres were extracted with a soxhlet extractor using 10% acetic acid in acetone for 48 h, thereafter the nanoshperes were ultrasonicated in 5 mL 20% acetic acid in acetone for 20 min and then centrifuged; the resulting supernatant was monitiored using HPLC. The ultrasonication step was repeated until no hypericin could be detected in the supernatant.
Control polymer nanospheres (NIP-NSs) were prepared under identical conditions but in the absence of hypericin.

Determination of Static Adsorption Capacity
To determine the static adsorption capacity, 5 mg of the nanospheres (MIP-NSHs or NIP-NSs) was placed into a 10 mL plastic centrifuge tube and mixed with 5 mL of the test molecule acetone solution (10.0 µM) to allow adsorption at 25 • C for 24 h. The concentration of the test molecule in the supernatant was measured with HPLC using a calibration curve (see Supplementary Materials Figure S6). The adsorption capacity (Q, µmol·g −1 ) was calculated according to Equation 1: where C 0 and C e represent the initial and equilibrium concentrations of the test molecule in acetone (µM), respectively, V (L) is the volume of the solution, and W (g) is the dry weight of the nanospheres. The neat adsorption capacity of MIP-NSHs over that of NIP-NSs, defined as the specific adsorption capacity of MIP-NSHs (Q s ) towards hypericin, is calculated according to Equation 2: where Q 1 and Q 2 are the static adsorption capacity of MIP-NSHs and NIP-NSs (µmol·g −1 ) towards hypericin, respectively.

Kinetic of Template Adsorption
5.0 mg of MIP-NSHs (or NIP-NSs) were mixed with a hypericin acetone solution (12.5 µM, 5 mL) in a 10-mL centrifuge tube. The tube was sealed and shaken in the dark at 25 • C for different time intervals (0.5, 1, 2, 4, 8, and 12 h, respectively). The concentrations of hypericin in the supernatant were determined by HPLC. The respective adsorptions were then calculated according to Equation 1.

Selectivity of MIP-NSHs and NIP-NSs for Hypericin
The binding selectivity of MIP-NSHs and NIP-NSs was evaluated by following the method reported [38]. Briefly, 5 mg of the MIP-NSHs or NIP-NSs were incubated respectively with 5 mL of hypericin, protohypericin, and emodin solution (12.5 µM in acetone) at 25 • C. After incubation under continuous shaking for 24 h, the amounts of hypericin, protohypericin, and emodin bound to the MIP-NSHs or NIP-NSs were measured, respectively. The binding selectivity of the NSs towards Polymers 2017, 9, 469 6 of 16 different molecules was compared using the "selectivity factor" (SF) and "imprinting factor" (IF) [39], which are determined as the ratio of distributions and can be calculated by the following equations: is the equilibrium concentration of hypericin for MIP (µM), C 0, MIP is the initial concentration of competitor for MIP (µM), and C e, MIP is the equilibrium concentration of competitor for MIP (µM).
where D MIP is the distribution of the test molecule for MIP (L·g −1 ), D NIP is the distribution of the test molecule for NIP (L·g −1 ), C 0, MIP is the initial concentration of the test molecule for MIP (µM), C e, MIP is the equilibrium concentration of the test molecule for MIP (µM), C 0, NIP is the initial concentration of the test molecule for NIP (µM), and C e, NIP is the initial concentration of the test molecule for NIP (µM).

The Reusability of MIP-NSHs
The reusability of MIP-NSHs was evaluated by an adsorption-extraction cycle experiment. One adsorption-extraction cycle consisted of loading the template and reaching equilibrium adsorption, followed by the extraction of the template. For the adsorption, 5 mg of the MIP-NSHs were incubated with 5 mL of hypericin solution (12.5 µM in acetone) at 25 • C. After incubation under continuous shaking for 8 h, the concentration of hypericin in the supernatant was measured by HPLC. The adsorption capacity was calculated according to Equation 1. The MIP-NSHs were subjected to the extraction procedure mentioned above, and then used for the next cycle.

Screening of Monomers and Crosslinkers
Previously, we have successfully prepared polymeric nanospheres via the alkyne−azide cycloaddition between either of the monomers and either of the cross-linkers, i.e., four "monomer + crosslinker" combinations (1 + 3 or 4 and 2 + 3 or 4) initiated by Cu(PPh 3 ) 3 Br [30]. However, due to the low polymer yield from the initiation system, a new photoinitiatior of DMPA was used for polymerization in this work. The four combinations were used for the preparation of imprinted polymers (denoted as P1, P2, P3, and P4 correspondingly) towards hypericin. The adsorption study showed that the MIP-NSHs obtained from 1 + 3 (Table 1, P1) gave the highest specific adsorption capacity (Q s = 1.91 µmol·g −1 ) among the four polymers, as shown in Table 1. Under the same conditions, Q s from 2 + 3 (Table 1, P2) was far lower at 0.33 µmol·g −1 . The reason may be that the steric hindrance of 2 is greater than that of 1 when forming the hydrogen bond between the N atom in pyridine ring and the phenolic hydroxyl groups in hypericin, which consequently affects the creation of the imprinting sites and thus impacts the recognition towards hypericin. In addition, using tetra-azide 4 instead of tri-azide 3 as the crosslinker did not benefit the adsorption of the resulting MIP-NSHs. For example, the Q s of P3 synthesized from 1 + 4 was much lower than that of P1. This might be because the polymer formed with 4 as crosslinker was denser than that with 3 as crosslinker, which make the removal and re-binding of hypericin from the polymer more difficult, and, as a result, lowered the adsorption of MIP-NSHs drastically. Thus, a tri-thiol crosslinker 5 was synthesized and combined with 1 to fabricate MIP-NSHs via thiol-yne click reaction (Table 1, P5). As expected, the Q s obtained was the best among the five tested combinations, which may be due to the absence of triazole rings and less rigid structure of P5. Therefore, the subsequent studies were based on the MIP-NSHs obtained from combination of monomer 1 and crosslinker 5.

Characterization of MIP-NSHs FTIR Analysis
To confirm that the thiol-yne click polymerization took place between 1 + 5 in the presence or absence of hypericin, both MIP-NSHs and NIP-NSs were analyzed using FTIR spectroscopy. The IR spectra are shown in Figure 2. In both scenarios, the characteristic absorption of C=O from 5 at 1735 cm −1 , 2103, 3274, and 2570 cm −1 ascribed to the absorptions of C≡C, C-H in alkynyl groups on 1 [30] and S-H on 5, respectively were not observed on the spectra of the polymers (MIP-NSHs and NIP-NSs). Moreover, the absorption band of C-O on hypericin at 1230 cm −1 [38] was found on the spectrum of MIP-NSHs. The results indicated that the click reaction of thiol-yne successfully occurred in both cases.

SEM and DLS Analysis
The polymers obtained via click polymerization between 1 + 5 were characterized with FESEM and DLS analyses. The results are shown in Figure 3, Table 2, and Figure S7. As shown in Figure 3, both of the imprinted and non-imprinted polymeric nanoparticles showed good spherical morphology. The difference between MIP-NSHs and NIP-NSs was that the surface of the former is much rougher, which can be clearly seen in the enlarged views (insets in Figure 3). DLS analysis showed that the average diameters of the MIP-NSHs and NIP-NSs were 677 and 497 nm, respectively; their polydispersity indexes were 1.137 and 0.994, respectively. In addition, the extraction process yielded an obvious change in the ζ potential of the MIP-NSHs, from 0.95 to −13.51 mV; in contrast, the ζ potential of NIP-NSs was less affected (see Table 2). This is in accordance with what has been reported previously in the literature [38]. The difference in the ζ potentials of MIP-NSHs before and after the extraction process is most likely due to the removal of hypericin molecules from the MIP-NSHs. All these results confirm that the MIP-NSHs can be prepared via thiol-yne click reaction of 1 + 5.

SEM and DLS Analysis
The polymers obtained via click polymerization between 1 + 5 were characterized with FESEM and DLS analyses. The results are shown in Figure 3, Table 2, and Figure S7 (see Supplementary Materials). As shown in Figure 3, both of the imprinted and non-imprinted polymeric nanoparticles showed good spherical morphology. The difference between MIP-NSHs and NIP-NSs was that the surface of the former is much rougher, which can be clearly seen in the enlarged views (insets in Figure 3). DLS analysis showed that the average diameters of the MIP-NSHs and NIP-NSs were 677 and 497 nm, respectively; their polydispersity indexes were 1.137 and 0.994, respectively. In addition, the extraction process yielded an obvious change in the ζ potential of the MIP-NSHs, from 0.95 to −13.51 mV; in contrast, the ζ potential of NIP-NSs was less affected (see Table 2). This is in accordance with what has been reported previously in the literature [38]. The difference in the ζ potentials of MIP-NSHs before and after the extraction process is most likely due to the removal of hypericin molecules from the MIP-NSHs. All these results confirm that the MIP-NSHs can be prepared via thiol-yne click reaction of 1 + 5.

SEM and DLS Analysis
The polymers obtained via click polymerization between 1 + 5 were characterized with FESEM and DLS analyses. The results are shown in Figure 3, Table 2, and Figure S7. As shown in Figure 3, both of the imprinted and non-imprinted polymeric nanoparticles showed good spherical morphology. The difference between MIP-NSHs and NIP-NSs was that the surface of the former is much rougher, which can be clearly seen in the enlarged views (insets in Figure 3). DLS analysis showed that the average diameters of the MIP-NSHs and NIP-NSs were 677 and 497 nm, respectively; their polydispersity indexes were 1.137 and 0.994, respectively. In addition, the extraction process yielded an obvious change in the ζ potential of the MIP-NSHs, from 0.95 to −13.51 mV; in contrast, the ζ potential of NIP-NSs was less affected (see Table 2). This is in accordance with what has been reported previously in the literature [38]. The difference in the ζ potentials of MIP-NSHs before and after the extraction process is most likely due to the removal of hypericin molecules from the MIP-NSHs. All these results confirm that the MIP-NSHs can be prepared via thiol-yne click reaction of 1 + 5.

BET Analysis
The average pore diameter, surface area and pore volume of the MIP-NSHs and NIP-NSs fabricated in this work were characterized with BET analysis. The results are summarized in Table 3 and Figure S8 (see Supplementary Materials). As can be seen, the average pore diameter and the pore volume of the MIP-NSHs were 52.839 nm and 0.033 cm 3 ·g −1 , respectively; which are much larger than those of NIP-NSs, indicating that the former has a porous structure, and the latter a compact one. The average pore diameter, surface area and pore volume of the MIP-NSHs were roughly 13, 2.5 and 8.4 times those of the NIP-NSs, respectively.

Optimization of Preparation Conditions for Specific Adsorption Capacity of MIP-NSHs
In order to obtain MIP-NSHs with the most optimal imprinting effect towards hypericin, the preparation conditions that affected Q s were screened and optimized; this included photoinitiator concentration, solvent composition, hypericin concentration, the equivalent ratio of 1 to 5, etc.
Effect of Photoinitiator Concentration on Q s As previously mentioned, DMPA was used as the photoinitiator in this work, which generated active free radicals upon exposure to UV light and induced monomer and crosslinker click polymerization. The concentration of DMPA affected not only the rate of polymerization, but also the imprinting effect of the resulting imprinted polymers. The effect of the concentration of DMPA on Q s was investigated by varying the concentration of DMPA in a range of 2.0-16 mg·mL −1 . As shown in Table 4, when the concentration of DMPA increased from 2.0 to 4.0 mg·mL −1 , Q s increased dramatically from 0.65 to 2.71 µmol·g −1 . However, further increase in the concentration led to the opposite effect with Q s decreasing to 0.57 µmol·g −1 when the concentration was increased to 16 mg·mL −1 . Hence, 4 mg·mL −1 was used as the optimal concentration of the photoinitiator for subsequent studies. Solvent plays an important role in the process of preparing MIP, especially in a non-covalent imprinting process. Proper selection of solvent may promote the formation of non-covalent adducts between functional monomers and templates and enhance the efficiency of imprinting [40]. Considering the solubility of hypericin, acetone was used in this study as a component of the solvent medium for polymerization. The effect of the ratio of acetone to acetonitrile on Q s of MIP-NSHs was studied by varying the ratio in a range of 4:0 to 1:3 (v/v). The results showed that the best ratio of acetone to acetonitrile was 3:1, where Q s reached 2.24 µmol·g −1 (Table 5); lower ratios led to a dramatic decrease in Q s , 1.47 µmol·g −1 with 1:1, and 0.29 µmol·g −1 with 1:3. The template concentration and ratio of monomer to crosslinker have a crucial influence on the property of MIPs [41]. They are important factors in dictating the number of binding sites and the rigidity of MIPs [42]. When there is sufficient amount of monomers for the pre-assembly of template and monomer, an increasing amount of template ensures a bigger number of binding sites and therefore a higher adsorption capacity, while an excessive crosslinker concentration (or too low ratio of monomer to crosslinker) leads to a highly crosslinked polymer, which impedes the template to transport in and out of the binding sites and thus decreases the adsorption capacity. This theory was further confirmed in this work. As shown in Table 6, when the concentration of hypericin increased from 0.625 to 2.5 mg·mL −1 , Polymers 2017, 9,469 10 of 16 Q s increased from 1.35 to 2.57 µmol·g −1 , however, further increase in the concentration of hypericin to 5.0 mg·mL −1 resulted in a clear decrease of Q s (see Table 6 P15). As the equivalent ratio of 1 to 5 decreased from 4:1 to 1:2, Q s increased at first, from 0.59 to a maximum value of 2.57 µmol·g −1 , and subsequently decrease dramatically to 0.45 µmol·g −1 when the ratio decreased to 1:2. Based on the results above, we conclude that the best combination of monomer/crosslinker for fabrication of MIP-NSHs in this work was 1 + 5. The MIP-NSHs with an optimal Q s value (2.57 µmol·g −1 ) can be prepared under the following conditions: the concentratration of hypericin: 5.0 mM; the concentration of monomer 1 and crosslinker 5 is 50 mM each; DMPA: 15.5 mM; solvent: acetone/acetonitrile in 3:1 (v/v), room temperature, UV light at 350 nm, 4 h. The MIP-NSH prepared under the optimal conditions described above, i.e. P14, were used for subsequent studies.

Kinetic of Template Adsorption
The equilibrium adsorption isotherms of MIP-NSHs and NIP-NSs for the binding of hypericin were investigated by batch adsorption experiments. As observed in Figure 4, the adsorption process of the MIP-NSHs displayed two phases: in the first phase, the adsorption amount increased quickly and reached to about one third of the total adsorption capacity during the first hour; and in the second phase, adsorption rate slowed down, and the equilibrium was reached at 8 h. The first phase can be attributed to the binding of hypericin to the recognition sites located on the surface of MIP-NSHs, which bound hypericin molecules at a fast rate. While the second phase can be attributed to the binding of hypericin to the internal binding sites of MIPNSHs, where the diffusion of hypericin from the surface to the inner part of the nanospheres resulted in the slow adsorption rate once the surface recognition sites became saturated. In addition, the equilibrium adsorption capacity of MIP-NSHs is higher than that of NIP-NSs, which indicates a fair imprinting effect. phase, adsorption rate slowed down, and the equilibrium was reached at 8 h. The first phase can be attributed to the binding of hypericin to the recognition sites located on the surface of MIP-NSHs, which bound hypericin molecules at a fast rate. While the second phase can be attributed to the binding of hypericin to the internal binding sites of MIPNSHs, where the diffusion of hypericin from the surface to the inner part of the nanospheres resulted in the slow adsorption rate once the surface recognition sites became saturated. In addition, the equilibrium adsorption capacity of MIP-NSHs is higher than that of NIP-NSs, which indicates a fair imprinting effect.

Affinity Analysis
The binding experiments were performed at different initial concentrations of hypericin, ranging from 0 to 50.0 μM, to compare the Q of the MIP-NSHs against that of NIP-NSs. Figure 5a shows the binding isotherms for hypericin on the MIP-NSHs and NIP-NSs. It can be seen that the Q of MIP-NSHs increased quickly in a linear relationship with the concentration of hypericin before Cs, a critical concentration of 11.5 μM. Thereafter, the increase of the Q slowed down and finally reached to a plateau. Furthermore, the amounts of substrate bound to the MIP-NSHs were more than that to the NIP-NSs, which was ascribed to the imprinting effect [43].
The adsorption mechanism study shows that the adsorption of MIP-NSHs towards hypericin fitted well with an extended Langmuir isotherm model (expressed by Equation 5, Figure 5b) [43,44], where R 2 is 0.9995, equilibrium constant Kd = 0.1843 μM, m is 1.8233, and Qmax is 6.07 μmol·g −1 . The results indicate the absorption may be described by monolayer sorption on a non-smooth surface.
Kd is the equilibrium constant (μM), Ce is the equilibrium concentration of hypericin (μM) in supernatant, and Qmax is the apparent maximum absorption capacity of binding sites (μmol·g −1 ).

Affinity Analysis
The binding experiments were performed at different initial concentrations of hypericin, ranging from 0 to 50.0 µM, to compare the Q of the MIP-NSHs against that of NIP-NSs. Figure 5a shows the binding isotherms for hypericin on the MIP-NSHs and NIP-NSs. It can be seen that the Q of MIP-NSHs increased quickly in a linear relationship with the concentration of hypericin before C s , a critical concentration of 11.5 µM. Thereafter, the increase of the Q slowed down and finally reached to a plateau. Furthermore, the amounts of substrate bound to the MIP-NSHs were more than that to the NIP-NSs, which was ascribed to the imprinting effect [43].
The adsorption mechanism study shows that the adsorption of MIP-NSHs towards hypericin fitted well with an extended Langmuir isotherm model (expressed by Equation 5, Figure 5b) [43,44], where R 2 is 0.9995, equilibrium constant K d = 0.1843 µM, m is 1.8233, and Q max is 6.07 µmol·g −1 . The results indicate the absorption may be described by monolayer sorption on a non-smooth surface.
K d is the equilibrium constant (µM), C e is the equilibrium concentration of hypericin (µM) in supernatant, and Q max is the apparent maximum absorption capacity of binding sites (µmol·g −1 ).  (6) Kd is the equilibrium constant (μM), Ce is the equilibrium concentration of hypericin (μM) in supernatant, and Qmax is the apparent maximum absorption capacity of binding sites (μmol·g −1 ).

Binding Selectivity
The binding selectivity of MIP-NSHs towards hypericin was examined by using a similar method reported previously [38], where protohypericin and emodin were used as competitors of the template (their structures are displayed in Figure 6a). MIP-NSHs or NIP-NSs were incubated respectively with

Binding Selectivity
The binding selectivity of MIP-NSHs towards hypericin was examined by using a similar method reported previously [38], where protohypericin and emodin were used as competitors of the template (their structures are displayed in Figure 6a). MIP-NSHs or NIP-NSs were incubated respectively with the same amount of hypericin, protohypericin, and emodin under the same conditions. The respective adsorption capacities of MIP-NSHs and NIP-NSs towards the three molecules are shown in Figure 6b. A higher adsorption for hypericin (6.03 µmol·g −1 ) was obtained with MIP-NSHs compared to protohypericin and emodin (2.36 and 1.30 µmol·g −1 , respectively). The binding selectivity of the NSs was evaluated with SF and IF, respectively. SF of MIP-NSHs towards protohypericin and emodin was 3.34 and 8.04, respectively; IF of MIP-NSHs towards hypericin, protohypericin, and emodin, was 2.44, 2.88, and 2.10, respectively (Table 7).

Binding Selectivity
The binding selectivity of MIP-NSHs towards hypericin was examined by using a similar method reported previously [38], where protohypericin and emodin were used as competitors of the template (their structures are displayed in Figure 6a). MIP-NSHs or NIP-NSs were incubated respectively with the same amount of hypericin, protohypericin, and emodin under the same conditions. The respective adsorption capacities of MIP-NSHs and NIP-NSs towards the three molecules are shown in Figure 6b. A higher adsorption for hypericin (6.03 μmol·g −1 ) was obtained with MIP-NSHs compared to protohypericin and emodin (2.36 and 1.30 μmol·g −1 , respectively). The binding selectivity of the NSs was evaluated with SF and IF, respectively. SF of MIP-NSHs towards protohypericin and emodin was 3.34 and 8.04, respectively; IF of MIP-NSHs towards hypericin, protohypericin, and emodin, was 2.44, 2.88, and 2.10, respectively (Table 7).

Reusability of MIP-NSHs
Imprinting materials as chemosensors are robust materials that are supposed to be reused many times, which is essential for reliable, economical and sustainable applications [45]. The evaluation on the reusability of the MIP-NSHs was performed and the results are shown in Figure 7. It can be seen that after five adsorption-extraction cycles, the adsorption capacity of the MIP-NSHs remained high, indicating good stability.

Reusability of MIP-NSHs
Imprinting materials as chemosensors are robust materials that are supposed to be reused many times, which is essential for reliable, economical and sustainable applications [45]. The evaluation on the reusability of the MIP-NSHs was performed and the results are shown in Figure 7. It can be seen that after five adsorption-extraction cycles, the adsorption capacity of the MIP-NSHs remained high, indicating good stability.

Conclusions
In conclusion, MIP-NSHs were successfully synthesized via thiol-yne click reaction using 3,5-diethynyl-pyridine as the monomer, tris(3-mercaptopropionate) as the crosslinker, and hypericin as the template. The click polymerization was completed in merely 4 h to produce the desired MIP-NSHs. The reaction conditions for adsorption capacity and selectivity towards hypericin were optimized, and the MIP-NSHs synthesized under the optimized conditions showed a high adsorption capacity (Q = 6.03 μmol·g −1 ) and fair selectivity towards hypericin. In addition, MIP-NSHs displayed good reusability up to at least five cycles. This work presents a straightforward method to fabricate molecularly imprinted nanospheres via thiol-yne click reaction.

Conclusions
In conclusion, MIP-NSHs were successfully synthesized via thiol-yne click reaction using 3,5-diethynyl-pyridine as the monomer, tris(3-mercaptopropionate) as the crosslinker, and hypericin as the template. The click polymerization was completed in merely 4 h to produce the desired MIP-NSHs. The reaction conditions for adsorption capacity and selectivity towards hypericin were optimized, and the MIP-NSHs synthesized under the optimized conditions showed a high adsorption capacity (Q = 6.03 µmol·g −1 ) and fair selectivity towards hypericin. In addition, MIP-NSHs displayed good reusability up to at least five cycles. This work presents a straightforward method to fabricate molecularly imprinted nanospheres via thiol-yne click reaction.