Construction of a chiral artificial enzyme used for enantioselective catalysis in live cells

Nanozymes as a newcomer in the artificial enzyme family have shown several advantages over natural enzymes such as their high stability in harsh environments, facile production on large scale, long storage time, low costs, and higher resistance to biodegradation. However, compared with natural enzymes, it is still a great challenge to design a nanozyme with high selectivity, especially high enantioselectivity. It is highly desirable and demanding to develop chiral nanozymes with high and on-demand enantioselectivity for practical applications. Herein, we present an unprecedented approach to construct chiral artificial peroxidase with ultrahigh enantioselectivity. Inspired by the structure of the natural enzyme horseradish peroxidase (HRP), we have constructed a series of stereoselective nanozymes (Fe3O4@Poly(AA)) by using the ferromagnetic nanoparticle (Fe3O4 NP) yolk as the catalytic core and amino acid-appended chiral polymer shell as the chiral selector. Among them, Fe3O4@Poly(d-Trp) exhibits the highest enantioselectivity. More intriguingly, their enantioselectivity will be readily reversed by replacing d-Trp with l-Trp. The selectivity factor is up to 5.38, even higher than that of HRP. Kinetic parameters, dialysis experiments, and molecular simulations together with activation energy reveal that the selectivity originates from the d-/l-Trp appended polymer shell, which can result in better affinity and catalytic activity to d-/l-tyrosinol. The artificial peroxidases have been used for asymmetric catalysis to prepare enantiopure d- or l-enantiomers. Besides, by using fluorescent labelled FITC-tyrosinolL and RhB-tyrosinolD, the artificial peroxidases can catalyze green or red fluorescent chiral tyrosinol to selectively label live yeast cells among yeast, S. aureus, E. coli and B. subtilis bacterial cells. This work opens a new avenue for better design of stereoselective artificial enzymes.


Introduction
The design and synthesis of articial enzymes or enzyme mimetics with high activity and selectivity have been the focus in the pursuit of alternatives for natural enzymes since the middle of the last century. 1 In the past 10 years, as a new generation of articial enzymes, nanomaterials with enzymemimetic activity, named nanozymes, have attracted great interest because of their many advantages over natural enzymes, including high stability in high temperature and harsh environments, facile production on a large scale, long storage time, low costs, and higher resistance to biodegradation. 2 Among them, the Fe 3 O 4 nanozyme is a classical nanomaterial with intrinsic peroxidase-like activity and was rst reported by Yan and coworkers in 2007. 3 Different types of nanozymes have been developed and successfully used for biosensing, 3a,4 cancer therapy, 5 combating bacteria and bio-lms, 6 detoxication, 7 and cytoprotection. 8 Although promising, compared with natural enzymes, rational design and synthesis of nanozymes are still challenging, 9,10 especially for the design of a chiral nanozyme with high selectivity. 11 Therefore, it is important and necessary to develop chiral nanozymes with high and on-demand enantioselectivity for biosynthesis and practical applications. 12 In natural enzymes, such as horseradish peroxidase (HRP, Scheme 1a, le), stereoselective reactions can be realized in the catalytic site (heme) with the assistance of proximally arrayed Lamino acid residues. 13 The utilization of chiral amino acids as chiral selectors has also been proved to be effective for the construction of chiral nanozymes. 11d,e However, the previously reported chiral nanozymes relied on the direct modication of amino acids on the surface of nanoparticles. The density of the chiral selector is relatively low, resulting in low selectivity. 11d,e Besides, part of the active sites are covered by direct modication, which can hinder their activity. 11d,e Thus, a new modication strategy is desirable for the design of chiral nanozymes. The yolk-shell structure has been conrmed to be powerful for the construction of catalysts and this structure can avoid direct covalent bonding of amino acids blocking the central catalytic site of nanozymes. 14 In addition, plenty of chiral amino acids can be introduced into the polymer shell to realize high efficiency chiral recognition. 15 Inspired by the natural enzyme and advantages of the yolk-shell structure, a yolk-shell chiral nanozyme is designed by using Fe 3 O 4 nanoparticle (NP) yolk as the catalytically active peroxidase-mimic and amino acidappended polymer shell as the selector for chiral recognition (Fe 3 O 4 @Poly(AA), where Poly(AA) refers to polyacrylate(amino acid), Scheme 1a, right). The Poly(AA) shell allows selective access of chiral tyrosinol to the surface of peroxidase-like cores, thus yielding enantioselective articial peroxidase. Scheme 1b illustrates well the synthesis process of Fe 3 O 4 @Poly(AA) via four steps, including preparing Fe 3 O 4 NPs, coating Fe 3 O 4 with a silica shell (Fe 3 O 4 @SiO 2 ), modifying Fe 3 O 4 @SiO 2 with a chiral polymer shell (Fe 3 O 4 @SiO 2 @Poly(AA)), and removing the silica shell to form yolk-shell Fe 3 O 4 @Poly(AA).

Results and discussion
Fe 3 O 4 NPs were rst prepared via a hydrothermal process by using citrate as a stabilizer. Then, Fe 3 O 4 NPs were coated with a SiO 2 shell and modied with 3-(trimethoxysilyl)propylmethacrylate (MPS) to introduce carbon-carbon double bonds. As shown in Fig. 1a and b,Fe 3 O 4 NPs with an average diameter of ca. 200 nm were homogeneously coated with a SiO 2 shell of about 25 nm thickness. Next, amino acid D-/L-tryptophan (Trp), D-/L-phenylalanine (Phe), D-/L-aspartic acid (Asp), and D-/L-histidine (His) appended polymer monomers were synthesized, respectively, and their corresponding 1 H nuclear magnetic resonance (NMR) spectra were shown in the ESI. † Then Fe 3 -O 4 @SiO 2 was coated with different amino acid appended polymers. By taking the L-Trp appended polymer as an example, the formation of the polymer shell on the Fe 3 O 4 @SiO 2 was rst characterized by transmission electron microscopy (TEM). Obviously, aer coating the polymer shell, the surface became rough (Fig. 1c). The thickness of the polymer shell was about 8.5 nm as determined by dynamic light scattering (DLS) at room temperature ( Fig. S1 †). The mid-layer silica of Fe 3 O 4 @SiO 2 @-Poly(L-Trp) was further selectively removed by etching in a highly concentrated NaOH solution to form yolk-shell structured Fe 3 O 4 @Poly(L-Trp) nanoparticles. As shown in Fig. 1d, a thin polymer coating around the surface of Fe 3 O 4 NPs was observed by TEM, which was consistent with the results of DLS determination (Fig. S1 †). The TEM elemental mapping images shown in Fig. 1e were further used to conrm the preparation of yolk-shell structured Fe 3 O 4 @Poly(L-Trp). Silica and polymer shells in Fe 3 O 4 @SiO 2 @Poly(L-Trp) were conrmed by the Scheme 1 (a) Representative structures of natural heme-containing peroxidase (left) and yolk-shell-structured artificial peroxidase (right). (b) Preparation procedures of the yolk-shell-structured artificial peroxidase. This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 11344-11350 | 11345 representative elements Si and N. As shown in Fig. 1f, NaOH selective etching led to the disappearance of Si and retention of N, further conrming the formation of yolk-shell Fe 3 O 4 @-Poly(L-Trp). The existence of the chiral polymer could also be supported by the obviously enhanced UV-vis absorption and circular dichroism (CD) spectra ( Fig. S2 and S3 †).The X-ray photoelectron spectroscopy (XPS) spectrum in Fig. S4 † showed minimal interaction between Fe 3 O 4 and the amino acid polymer, further conrming the successful synthesis of the yolkshell Fe 3 O 4 @Poly(L-/D-Trp). 3,3,5,5-tetramethylbenzidine (TMB) was used to measure the catalytic activity of Fe 3 O 4 @Poly(AA) in the presence of H 2 O 2 . All the catalytic activities of Fe 3 O 4 @-Poly(AA) were found to be similar to that of bare Fe 3 O 4 NPs, as reported previously. 3 Thus, yolk-shell Fe 3 O 4 @Poly(AA) nanoparticles still exhibit peroxidase activity (Fig. S5 †). Notably, the core-shell-structured Fe 3 O 4 @SiO 2 @Poly(AA) showed no catalytic activity due to the blocking of the catalytically active site (data not shown), suggesting the necessity of a yolk-shell structure. Taking Fe 3 O 4 @Poly(L-Trp) as an example, it had peroxidase activity in a wide pH range ( Fig. S5 †).
In the following, the catalytic activities of Fe 3 O 4 @Poly(AA) toward a specic chiral substrate (L-/D-tyrosinol) were further studied by monitoring the time-dependent absorbance at 320 nm (Fig. 2a). The initial rates of the oxidation of L-/D-tyrosinol enantiomers could reveal the enantioselectivity of Fe 3 -O 4 @Poly(AA). Fig. 2b showed the initial rates for the oxidation of L-and D-tyrosinol with Fe 3 O 4 @Poly(L-Trp), Fe 3 O 4 @Poly(D-Trp) and bare Fe 3 O 4 , respectively. Fe 3 O 4 @Poly(D-Trp) showed a better catalytic activity for D-tyrosinol than L-tyrosinol. The ratio calculated using the initial rates of oxidation of the D enantiomer and L enantiomer for Fe 3 O 4 @Poly(D-Trp) was 2.38. Interestingly, the enantioselectivity was reversed completely by using Fe 3 O 4 @Poly(L-Trp) with a ratio of 0.37, which was consistent with the rst principles of stereochemistry. 16 This indicated that the enantioselectivity of our designed nanozyme could be reversed readily by changing the chirality of the polymer shell. Similarly, the oxidation of D-/L-tyrosinol enantiomers by a series of Fe 3 O 4 @Poly(AA) was also investigated and the corresponding results were shown in Fig. 2c. As shown in Fig. 2c, Fe 3 O 4 @Poly(Phe) and Fe 3 O 4 @Poly(Asp) showed moderate enantioselectivity, while Fe 3 O 4 @Poly(His) showed poor enantioselectivity. In contrast, bare Fe 3 O 4 did not show any stereoselectivity for oxidizing L-or D-tyrosinol. To better understand the stereoselectivity of Fe 3 O 4 @-Poly(Trp) towards tyrosinol enantiomers, we analyzed and compared the kinetic parameters based on the Michaelis-Menten model according to the saturation curves in Fig. S6. † The results were summarized in Table S1. † The smaller the K M , the higher the binding affinity between enzymes and substrates. The catalytic number (k cat ) indicates the ability of enzymes to catalyze one certain substrate. k cat /K M is usually applied for describing the catalytic efficiency of the enzyme. For Fe 3 O 4 @-Poly(D-Trp), k cat values for D-and L-tyrosinol were (16.44 AE 0.40) Â10 3 s À1 and (10.78 AE 1.27)Â10 3 s À1 , respectively, indicating the higher catalytic ability of Fe 3 O 4 @Poly(D-Trp) for oxidizing Dtyrosinol rather than L-tyrosinol. Moreover, a lower K M value was observed for D-tyrosinol than for L-tyrosinol, implying stronger binding affinity of Fe 3 O 4 @Poly(D-Trp) for D-tyrosinol. The higher catalytic activity and binding affinity to D-tyrosinol resulted in higher catalytic efficiency toward D-tyrosinol. As decided by k cat /K M values, Fe 3 O 4 @Poly(D-Trp) was 5.38 times more active for D-tyrosinol than for L-tyrosinol. As for Fe 3 O 4 @-Poly(L-Trp), the preference was reversed to L-tyrosinol due to the better catalytic ability and binding affinity to L-tyrosinol. Thus, the selectivity of the yolk-shell nanozyme originated from the higher activity and affinity to D-/L-tyrosinol and the preference was related to the chirality of the polymer. The selectivity factor of the yolk-shell nanozyme was higher than those of many chiral nanozymes reported before, such as graphene oxidebased peroxidase, 11a gold nanoparticle-based glucose oxidase, 11b gold nanoparticle-based phosphorylase, 11c ceria nanoparticle-based oxidase 11d and gold nanoparticle-based peroxidase 11e (Table S2 †). In addition, in the oxidation of tyrosinol enantiomers, the Fe 3 O 4 @Poly(D-Trp) had a higher selectivity factor than HRP, which was reported to be 4.77 according to a previous report (Table S2 †). 13c The improvement of enantioselectivity could be attributed to the strategies of the surface modications. 9b,17 Compared with the previously reported chiral nanozymes, in which the catalytic sites were partly covered by the direct modication of the chiral selector even though the density of the chiral selector was relatively low, our designed yolk-shell structure allowed highly dense modication of amino acids as the chiral selector but not direct covalent bonding to the central catalytic site. This enhanced the selectivity without blocking the catalytic site of the nanozyme. 15 This design endowed the articial peroxidase with high and ondemand enantioselectivity.
The selective binding affinity and catalytic activity of Fe 3 -O 4 @Poly(AA) toward L-/D-tyrosinol were further conrmed. We rst conducted dialysis experiments to reveal the binding preference. 18 Fe 3 O 4 @Poly(AA) in the dialysis tube was dialyzed against racemic mixtures of tyrosinol, and circular dichroism was used to monitor the dialysate for enrichment of the enantiomer with weaker interaction with Fe 3 O 4 @Poly(AA). As shown in Fig. S7, † the dialysate for bare Fe 3 O 4 was racemic, indicating no binding preference to tyrosinol enantiomers. The dialysate for Fe 3 O 4 @Poly(L-Trp) and Fe 3 O 4 @Poly(D-Trp) was rich in Dtyrosinol and L-tyrosinol, respectively, which obviously indicated the higher affinity of Fe 3 O 4 @Poly(L-Trp) to L-tyrosinol and Fe 3 O 4 @Poly(D-Trp) to D-tyrosinol. Here we proposed a possible catalytic process: D-/L-tyrosinol rst diffused through the Poly(D-/L-Trp) shell, then reacted with the hydroxyl radicals, which were formed from the Fe 3 O 4 NP-catalyzed decomposition of hydrogen peroxide. The obtained phenoxy radical could then couple with another D-/L-tyrosinol to give the nal products. We supposed that the selectivity should be more relevant to the diffusion process of D-/L-tyrosinol since the bare Fe 3 O 4 NPs displayed no selectivity to the tyrosinol enantiomers. The diffusion of L-/D-tyrosinol into the inner catalytic core through the Poly(D-/L-Trp) shell could be divided into three processes: adsorption, transport and desorption. We then examined the role of L-/D-tyrosinol adsorption/desorption using a patch of Poly(D-/L-Trp) shell models by molecular simulation methods (Fig. S8 †). The amino acid-appended polymer shell could selectively transport L-/D-tyrosinol from the outer solution into the inner catalytic yolk. Therefore, the adsorption of L-/D-tyrosinol on the outer surface of the Poly(D-/L-Trp) shell and the desorption from the inner surface of the Poly(D-/L-Trp) shell could give information about the binding preference. In order to clearly demonstrate this fact, we investigated the top 3 binding structures of L-/D-tyrosinol with the outer and inner surface of Poly(D-/L-Trp) (shown in Fig. S9 †), respectively. The detailed illustration in Fig. 3 depicted the best binding mode of L-/D-tyrosinol on the outer surface of the Poly(D-/L-Trp) shell. We also examined their corresponding average adsorption free energy (AFE) on the outer surface and desorption free energy (DFE) from the inner surface of Poly(D-/L-Trp) (listed in Table  S3 †) based on the top 3 binding models. It was found that the AFE of D-tyrosinol on the outer surface of Poly(D-Trp) was much lower than that of L-tyrosinol by 2.9 kcal mol À1 , indicating that Poly(D-Trp) was more prone to adsorb D-tyrosinol, not L-tyrosinol. Similarly, Poly(L-Trp) was more favourable for the adsorption of L-tyrosinol with a free energy difference of 1.8 kcal mol À1 compared to D-tyrosinol. More interestingly, we also noted that the desorption of D-tyrosinol from Poly(D-Trp) and L-tyrosinol from Poly(L-Trp) was easier than the opposite conguration because of the lower binding free energy. Nevertheless, it was also found that the differences of adsorption free energy were distinctly larger than those of desorption free energy, indicating that the adsorption process played an important role in the enantioselectivity of Poly(D-/L-Trp). Furthermore, we also decomposed the binding free energy into individual energy components to shed light on the dominant interaction for driving D-tyrosinol and L-tyrosinol binding to Poly(D-/L-Trp), as presented in Table S4. † We found that van der Waals interactions provided a major contribution to the difference of adsorption free energy of D-tyrosinol and L-tyrosinol on the outer surface of Poly(D-Trp) and Poly(L-Trp). This indicated that van der Waals interactions were closely correlated with the enantioselectivity of Fe 3 O 4 @Poly(D-/L-Trp). Electrostatic interactions, determined by the hydrogen bond interactions, were favourable for the adsorption of D-tyrosinol on the outer surface of Poly(D-Trp), while it showed unfavourable contributions to the adsorption of L-tyrosinol on the outer surface of Poly(L-Trp). The results of molecular simulation were consistent with dialysis experiments, and demonstrated that Poly(D-Trp) and Poly(L-Trp) contributed to the binding preference of Fe 3 O 4 @Poly(D-/L-Trp) to D-/L-tyrosinol. It is worth mentioning that the transport dynamics of L-/D-tyrosinol within the Poly(D-/L-Trp) shell should also play an important role in the catalytic enantioselectivity, which will be studied in the future. Then, the activation energy (E a ) was calculated according to the Arrhenius equation 19 (Fig. S10 †). For Fe 3 O 4 @Poly(D-Trp), the E a values for D-tyrosinol and L-tyrosinol were estimated to be 65.1 AE 2.3 kJ mol À1 and 86.6 AE 2.4 kJ mol À1 , respectively. It was obvious that in the presence of Fe 3 O 4 @Poly(D-Trp), the energy barrier for the oxidation of D-tyrosinol was lower. In contrast, Fe 3 O 4 @Poly(L-Trp) exhibited a lower energy barrier for the oxidation of Ltyrosinol. The results conrmed undoubtedly the higher catalytic activity of Fe 3 O 4 @Poly(D-Trp) to D-tryosinol and Fe 3 O 4 @-Poly(L-Trp) to L-tyrosinol, which was consistent with the kinetic parameters.
The performance of Fe 3 O 4 @Poly(L-/D-Trp) in enantioselective catalysis was further studied by high-performance liquid chromatography (HPLC). As shown in Fig. S11, † in the presence of Fe 3 O 4 @Poly(L-Trp), L-tyrosinol was almost consumed to generate the product with an 89% yield. The main product was found to be L-dityrosinol according to the 1 H NMR spectra in Fig. S12a. † In contrast, the yield for the reaction of D-tyrosinol was just 32%. For Fe 3 O 4 @Poly(D-Trp), D-tyrosinol was almost transformed to D-dityrosinol with a 92% yield as shown in This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 11344-11350 | 11347 Fig. S11d and S12b, † while the yield for the reaction of L-tyrosinol was only 29%. The results displayed the catalytic selectivity of our nanozymes toward L-/D-tyrosinol. The chiral nanozyme was then applied for the oxidation of racemic tyrosinol. As shown in Fig. S13, † with Fe 3 O 4 @Poly(L-Trp), L-tyrosinol was almost consumed while D-tyrosinol was in excess, indicating the higher conversion of L-tyrosinol. As for Fe 3 O 4 @Poly(D-Trp), D-tyrosinol was almost consumed while L-tyrosinol was in excess, indicating the higher conversion of D-tyrosinol. The oxidation of the racemic substrate further proved that the oxidation of tyrosinol by Fe 3 O 4 @Poly(L-/D-Trp) was enantioselective. Besides, the enantiomeric excess was up to 99% toward D-tyrosinol with the catalysis of Fe 3 O 4 @Poly(L-Trp). This meant that enantiopure D-tyrosinol was prepared. Similarly, enantiopure L-tyrosinol was prepared from racemate tyrosinol with the catalysis of Fe 3 O 4 @Poly(D-Trp). Signicantly, the results suggested the potential ability of our articial peroxidase for preparation of optically pure compounds.
Then, we veried whether the catalytic activity and selectivity toward L-/D-tyrosinol of our designed nanozymes can be realized in living cells. Live/dead cell staining was rst performed. As shown in Fig. S14, † the nanozyme and H 2 O 2 showed minimal inuence on the cell viability. Klibanov et al. demonstrated that reactive oxygen species can catalyze tyrosinol to form free phenol radicals, subsequently covalently linked to the tyrosine residues on the surface of yeast cells. 20 For visualizing these processes, we modied the L-/D-tyrosinol with FITC and Rhodamine B (RhB), marked as FITC-tyrosinol L and RhBtyrosinol D . Yeast cells were treated with FITC-tyrosinol L and RhB-tyrosinol D in the presence of bare Fe 3 O 4 , Fe 3 O 4 @Poly(L-Trp) or Fe 3 O 4 @Poly(D-Trp), respectively (Scheme S1 †). As shown in Fig. 4, for the control group, negligible uorescence was observed in the yeast cells. For the bare Fe 3 O 4 group, both green and red uorescence were observed, which suggested that bare Fe 3 O 4 NPs had no enantioselectivity towards tyrosinol. For the Fe 3 O 4 @Poly(L-Trp) group, the green uorescence was much stronger than the red uorescence, indicating the high enantioselectivity of Fe 3 O 4 @Poly(L-Trp) toward L-tyrosinol. For the Fe 3 O 4 @Poly(D-Trp) group, strong red uorescence was observed on the surface of yeast cells, indicating the high enantioselectivity of Fe 3 O 4 @Poly(D-Trp) towards D-tyrosinol. Flow cytometry ( Fig. 4b) also showed similar results, further supporting the confocal uorescence data.
We also attempted to label bacterial cells, such as S.aureus, E.coli and B.subtilis with FITC-tyrosinol L /RhB-tyrosinol D by using the same assay as we did for yeast cells. As shown in Fig. S15-20, † neither green emission nor red emission was observed in all groups including the control, bare Fe 3 O 4 , Fe 3 -O 4 @Poly(L-Trp) and Fe 3 O 4 @Poly(D-Trp). These results indicated that the surface of S.aureus, E.coli and B.subtilis cells were difficult to label with a tyrosinol-based uorescent agent. Next four different microbial cells (yeast, S.aureus, E.coli and B.subtilis) were treated with Fe 3 O 4 @poly(L-Trp), H 2 O 2 , FITCtyrosinol L and RhB-tyrosinol D to demonstrate the specicity of the labelling process towards yeast. As shown in Fig. 5, only yeast cells showed green uorescence among these cells because of the abundant tyrosine residues on the surface of yeast cells. 20 The different labelling results on S.aureus, E.coli and B.subtilis might be due to the different structures and chemical compositions of their cell walls. The specic feature of the Gram-positive bacterial cell wall (S. aureus) is that it is composed of teichoic/lipoteichoic acids. The characteristic components of the Gram-negative bacterial cell wall (E.coli and B.subtilis) are lipopolysaccharides. 21 Fewer tyrosine residues exist on the bacterial cell wall compared with the yeast cells. Thus fewer uorescent agents could be labelled on their surface.

Conclusions
In summary, we present an unprecedented approach to integrate the active yolk articial peroxidase Fe 3 O 4 NP and D-or Lamino acid-appended chiral selective polymer shell for construction of a series of yolk-shell structured articial peroxidases. Different from the chiral nanozymes with direct  surface modication, the chiral articial peroxidases exhibit high enantioselectivity. Among them, Fe 3 O 4 @Poly(D-Trp) shows the highest selectivity factor of 5.38, which is even higher than that of HRP. The enantioselectivity is readily reversed by replacing D-Trp with L-Trp. Kinetic parameters, dialysis experiments, and molecular simulations together with activation energy reveal that the D-/L-Trp appended polymer shell results in better affinity and catalytic activity to D-/L-tyrosinol, thus leading to the selectivity of Fe 3 O 4 @Poly(D-Trp) to D-tryosinol and Fe 3 -O 4 @Poly(L-Trp) to L-tryosinol, respectively. These articial peroxidases could be used for preparing enantiopure D-or Lenantiomers. Besides, confocal uorescence microscopy and ow cytometry analysis have demonstrated that uorescent dye-modied chiral tyrosinol can selectively label yeast cells among S.aureus, yeast, E.coli and B.subtilis in the presence of our designed nanozyme. This work would promote rational design and synthesis of stereoselective articial enzymes.

Conflicts of interest
There are no conicts to declare.