Rational Design and Synthesis of an Artificial Enzyme for S_N2 Reactions through Micellar Imprinting

Abstract

Rational design of catalysts with enzyme-like properties is an elusive goal of chemists despite tremendous interest. Molecular imprinting inside surfactant micelles, followed by postmodification, creates a tailored active site in a water-soluble polymeric “artificial enzyme” for the benzylation of 4-nitrophenol. The reaction happens under neutral conditions, with excellent substrate selectivity. Similar to many enzymes, electrostatics play vital roles in the catalysis and can be tuned rationally through different bases introduced in the active site.

Graphical Abstract

Efficient catalysis under mild conditions and extraordinary substrate specificity of enzymes have captivated the imagination of researchers for generations.^1–6 If substrate-tailored active sites with accurately positioned catalytic groups can be constructed rationally, there is no fundamental reason why synthetic catalysts cannot work like enzymes because they are governed by the same fundamental principles.^7,8 In reality, however, “artificial enzymes” made by chemists rarely achieve enzymatic activity even though rational design of enzyme-like catalysts has been a long-standing goal for decades.^9–12

Despite tremendous progress in synthetic chemistry, chemists are generally better at making molecules rather than a functionalized nanospace tailored for the substrate(s)—i.e., the active site. Molecular imprinting has the potential to overcome this difficulty.^13–15 By co-polymerizing a mixture of template molecules, functional monomers (FMs), and cross-linkers, researchers construct template-complementary imprinted sites inside a polymer network. If equipped with catalytic groups, these imprinted sites can catalyze a range of reactions.^16–22 Although widely used for separation, sensing, and other analytical applications, molecularly imprinted polymers (MIPs) are less developed for catalysis due to several notable challenges with traditional MIPs including a heterogeneous distribution of binding sites and slow mass transfer in a heavily cross-linked polymer network.

In this work, we demonstrate that it is possible to rationally design a nanoparticle catalyst to catalyze alkylation of phenol. There is no natural enzyme that can catalyze such reactions, yet the synthetic catalysts developed have all the characteristics of an enzyme including reaction near physiological conditions, Michaelis–Menten kinetics, and exquisite substrate specificity.

The general preparation for our molecularly imprinted nanoparticles (MINPs) involves solubilization of a template molecule in the mixed micelle of cross-linkable surfactant 1, together with a radical cross-linker (divinylbenzene or DVB) and a hydrophobic radical initiator (Scheme 1a).²³ Surface-cross-linking of the micelle with a diazide (3) by the highly efficient click chemistry (step a), followed by surface-functionalization of the micelle by monoazide 3 (step b), yields an organic nanoparticle with the template still bound. Free radical polymerization initiated photolytically or thermally crosslinks the micellar core, around the template to form a template-complementary imprinted site inside the doubly cross-linked micelle (step c). Removal of the template vacates the imprinted site.

SCHEME 1.

(a) General procedure for preparing molecularly imprinted nanoparticles (MINPs). (b) Preparation of catalytic MINP for 4NP benzylation. The surface ligands (3) are omitted for clarity (see Scheme S4 for a more detailed drawing).

Our model S_N2 reaction is the alkylation of 4-nitrophenol (4NP) by benzyl chloride (BnCl). Although the reaction happens readily under basic conditions (to generate the phenoxide as a stronger nucleophile), simply mixing the two gives no reaction. To catalyze the reaction under neutral conditions, we designed and synthesized template 4 (Scheme 1b). It is color-coded for the different purposes of each substructure: the red benzoyl group mimics the electrophile (BnCl), the magenta moiety is similar to the nucleophile (4NP) in size and shape, and the blue substructure contains a photocleavable ortho-nitrobenzyl ester.

Photolysis of MINP(4) produces MINP(4)-CO₂H, with the template cleaved and a carboxylic acid left in the imprinted site.²³ The carboxylic acid is activated by 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDCI) and then amidated²³ with amine 5 that is similar in size and shape to the part of the template removed by the photolysis. Our expectation is that geometrical matching will help the pyridyl group of MINP(4+5) reside next to the bound 4NP, for its deprotonation and facilitated nucleophilic attack on the BnCl molecule in the active site (Scheme 1b, lower left structure).

Preparation of the MINPs and their characterizations are reported in the Supporting Information. Characteristic changes in the ¹H NMR spectrum (Figures S1, e.g., disappearance of the alkene protons) and in the particle size measured by dynamic light scattering (DLS, Figures S2–4) allowed us to monitor the cross-linking at different stages.

Figure 1a provides the initial evidence for MINP(4+5) as an artificial “nucleophilic substitutase”. In the presence of BnCl and MINP(4+5), the absorbance of 4NP at 400 nm decreases rapidly, whereas it stays nearly constant in case of the nonimprinted nanoparticles (NINP) prepared without the template. Similar to an enzyme, the reaction follows Michaelis–Menten kinetics (Figure 1b), affording a k_cat of 0.25 min⁻¹, a K_m of 47.4 ± 3.3 μM, and a catalytic efficiency (k_cat/K_m) of 89 M⁻¹s⁻¹. Consistent with the S_N2 reaction, benzyl 4-nitrophenyl ether was clearly identified (by comparison to the ¹H NMR spectrum of an authentic sample) when 4NP and BnCl was treated with catalytic amounts of the MINP (Figure S6). The product is larger and more hydrophobic than the reactants, thus having a stronger hydrophobic driving force to stay in the active site. This will be offset by the loss of hydrogen bond between the active site base and phenol/phenoxide. In the end, our catalyst still has a strong ability to turn over substrates, with a total turnover number (TON) of 167 measured at 38 min (Figure S5).

FIGURE 1.

(a) Absorbance at 400 nm of a mixture of 4NP and BnCl in the presence of MINP(4+5) (black curve) and NINP (blue curve) in a 25 mM HEPES buffer (pH 7.0) at 40 °C. [4NP] =50 μM. [BnCl] = 1500 μM. [MINP] = [NINP] = 8.0 μM. (b) Michaelis-Menten plot for a mixture [4NP] & [BnCl] in the presence of MINP(4+5) in a 25 mM HEPES buffer (pH 7.0) at 40 °C.

Since the background reaction for the benzylation of 4NP is practically zero (Figure 1a, blue curve) and the reaction only occurs inside the MINP(4+5) active site, we expect a strong selectivity for the substrates. Figure 2 shows that both the nucleophile (phenol) and the electrophile (benzyl chloride or bromide) need to fit into the active site for the catalyzed reaction to occur (see Figures S6–10 for the corresponding ¹H NMR spectra). Whenever the nitro group on the phenol is in the wrong position (ortho) or an extra nitro exists (ortho and para), alkylation of the phenol stops completely, due to the inability of the substrate to enter the catalytic active site. The same is true with the electrophile, with the alkylation absent when an isopropyl is introduced at the para position or methoxy at the meta position of the benzyl bromide/chloride.

FIGURE 2.

Selectivity of MINP(4+5) in the catalytic S_N2 reactions after 5 h in a 25 mM HEPES buffer (pH 7.0) at 40 °C. [Phenol] = 2.0 mM. [BnCl or BnBr] = 8.0 mM. [MINP]= 0.24 mM. Yields were determined by ¹H-NMR spectroscopy with 2.0 mM CH₂Br₂ as the internal standard.

To further illustrate the substrate specificity, we performed two competitive reactions. In Figure 3a, 2-nitrophenol (2NP) is first added to a mixture of BnCl and MINP(4+5). The unchanged UV absorbance at 400 nm indicates that 2NP (which has a smaller extinction coefficient than 4NP at 400 nm) is not alkylated under the reaction condition. As soon as 4NP is added at 10 min, the large increase of absorbance (as a result of more phenol added to the solution) is followed by a quick decrease, due to the 4NP benzylation. When the order of addition is reversed, Figure 3b shows that 4NP is consumed as soon as it is added to the reaction mixture. When 2NP is added at 25 min, the decrease in absorbance after the jump is nearly absent—the very small decrease should come from slow 4NP benzylation after 25 min.

FIGURE 3.

Competitive SN2 reaction between 2NP and 4NP catalyzed by MINP(4+5) in a 25 mM HEPES buffer (pH 7.0) as measured by the absorbance at 400 nm. [2NP] = [4NP] = 50 μM. [BnCl] = 1500 μM. [MINP] = 8.0 μM.

Figure 4a shows the pH profile of 4NP benzylation under different conditions. The reaction in buffer and in the presence of NINP are negligible over pH 6–8.5. In contrast, MINP(4+5) accelerates the benzylation greatly and, interestingly, displays a maximum at pH 7 (green curve). A slower reaction below pH 7 is not surprising for the 4NP benzylation, because the pK_a of 4NP is 7.15²⁴ and a protonated phenol has a much lower nucleophilicity than the deprotonated phenoxide. If the above is the only important factor in the catalyzed reactions, however, it is difficult to explain why the reaction slows down also above pH 7.

FIGURE 4.

(a) pH profiles of the reaction between 4NP and BnCl at 40 °C under different catalytic conditions. [4NP] = 50 μM. [BnCl] = 1500 μM. [MINP] = 8.0 μM. Buffer: MES for pH 6–6.5 and HEPES for pH 7.0–8.5. (b) Proposed mechanisms for the 4NP benzylation in the MINP active site.

For an S_N2 reaction, the rate-determining step (RDS) is the nucleophilic attack on the electrophile. There are two potential ways for our MINP catalyst to achieve the catalysis: in path A, the base acts as a general base to deprotonate the phenol, which in a concerted fashion attacks benzyl chloride (shown by the blue curved arrows in Figure 4b); in path B, the catalyst at the resting state uses the protonated base to assist the deprotonation of the phenol by electrostatic interactions, because a neighboring positive exerts strong repulsion to a (positively charged) proton and helps deprotonation of acids (in enzymes and synthetic systems).^25–27 This allows a phenoxide, a stronger nucleophile, to be generated in the active site, which attacks the nearby benzyl chloride in the catalyzed reaction (shown by the red curved arrows in Figure 4b). Both mechanisms require deprotonation of the phenol for the S_N2 reaction²⁸ but the timing is different for the two mechanisms.

The unusual pH profile of MINP(4+5) can be explained by mechanism B in a trade-off between two conflicting factors. At lower pHs, the pyridinium dominates in the active site and can assist the deprotonation of 4NP by its electrostatic repulsion. However, a lower solution pH makes the phenol deprotonation less favorable, slowing down the reaction in consequence. The opposite is true at higher pHs—whereas the 4NP deprotonation becomes more favorable, the electrostatic assistance from the pyridinium is lost as the neutral pyridine becomes the dominant species in the active site. At the end, pH 7 probably happens to be the optimal pH for the catalysis to occur in MINP(4+5).

There are two additional lines of evidence to support mechanism B. First, when the active site pyridine is replaced with a more basic guanidine, MINP(4+6) displays a higher activity than MINP(4+5) above ≥ pH 7 and, more importantly, its activity is nearly constant over pH 7–8.5 (Figure 4, black curve). Meanwhile, the catalytic efficiency (k_cat/K_m) for the 4NP benzylation increases from 89 M⁻¹s⁻¹ for MINP(4+5) (Figure 1b) to 133 M⁻¹s⁻¹ for MINP(4+6) (Figure S11). Guanidine being a strong base (pK_a >13) is expected to be protonated throughout this pH range. Thus, the electrostatic assistance of the 4NP deprotonation is always present with this catalyst under our experimental conditions. Under this scenario, the phenoxide can be generated fully ≥ pH 7 and gives the same high reaction rate. Dominance of the catalyzed reaction is evident from (a) the lack of reactivity in buffer or in the presence of NINP under the same conditions (Figure 4a), and (b) the nearly constant reactivity in the MINP(4+6)-catalyzed reaction.

Second, path A involves deprotonation of the phenol in the RDS but the proton transfer is already complete before the RDS in path B. Solvent kinetic isotope effects (SKIEs) can distinguish the two situations: without the O-H bond cleavage in the RDS, k_H2O/k_D2O is expected to be close to unity but the number increases to 2–3 for a typical general base catalysis.^29–32 In our hands, MINP(4+5) and MINP(4+6) display an SKIE of 1.03–1.08 (Table S1), supporting path B as the dominant catalytic mechanism. For the measurements, the pD value was determined by adding 0.4 to the pH meter reading, since water and D₂O have different dissociation constants.³⁰

In summary, a rational design of an artificial “nucleophilic substitutase” is reported in this work. Similar to many enzymes,⁶ electrostatic interactions are vital to the facilitated deprotonation of the phenol substrate to generate a stronger nucleophile near the bound electrophile. Although similar S_N2 reactions can happen readily in solution without a catalyst, the uncatalyzed reactions require equivalent amounts of base and will not be able to differentiate closely related phenols as our MINP catalysts.

It is encouraging that MINP-based catalysts display some of the key features of enzymes including efficient catalysis under mild conditions and a strong substrate selectivity. We believe the key reason for the success is the high fidelity of micellar imprinting: the imprint/nonimprint ratio in binding (i.e., imprinting factor) reaches up to 10,000 for some templates (peptides),³³ and MINPs can distinguish the addition,³⁴ removal,³⁴ and shift³⁵ of a single methyl (or methylene) group in the structure of the guest molecule during binding. This essentially enables us to quickly construct functionalized active sites with accurately positioned catalytic groups. Other types of enzyme-like catalysts should become possible as additional designs are validated.^23,36

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.