Turning Enzyme Models into Model Enzymes in a Substrate-Tailored, Desolvated Active Site

Abstract

Once taken out of the active sites, the same functional groups used by enzymes for catalysis tend to lose their “magical” catalytic power in small-molecule enzyme models. We report a small-molecule enzyme model that activates a benzylic alcohol by a nearby amine for the nucleophilic attack of an activated ester. Only when the two groups are placed inside a substrate-tailored hydrophobic pocket can they display catalytic turnovers and even become reactive enough to hydrolyze amides catalytically near physiological conditions, a long-standing goal for synthetic protease mimics. These results suggest that the large gap between the innumerable catalytically incompetent small-molecule enzyme models made by chemists and true enzyme-like catalysts could be bridged by environmental engineering, which in this work enables a simple combination of a tertiary amine and an alcohol to replicate the catalytic properties of serine protease in hydrolyzing aryl amides with substrate specificity.

Enzymes routinely employ functional groups with medium or low intrinsic activities for highly challenging catalytic reactions. To understand enzymatic catalysis and ultimately be able to make synthetic catalysts with enzyme-like efficiency and specificity, generations of chemists have designed and synthesized innumerable small-molecule enzyme models bearing certain features of enzyme active sites. − Although these molecular models provided important mechanistic insights into enzymatic catalysis, the general finding is that once taken out of the enzyme active site, the same catalytic groups almost always lose their catalytic power.

A well-studied example is serine protease, which uses a neighboring histidine–aspartate pair to activate an active site serine for a nucleophilic attack on amide, an otherwise impossible task for a weak nucleophile (i.e., alcohol) under physiological conditions. Chemists have prepared numerous small-molecule models of serine protease, spanning several decades of time. − Although activities toward activated estersoccasionally even nonactivated estershave been observed with these − and related polymeric/self-assembled systems, − serine-protease-like catalytic hydrolysis of amide bonds has not been achieved.

We designed molecule 1a as a molecular model of serine protease, envisioning that its ortho tertiary amine could activate the benzylic hydroxyl via an intramolecular hydrogen bond (Figure ). The benzylic alcohol is indeed found to abstract the acetyl from p-nitrophenyl acetate (2a) in a methanol solution, although not catalytically. For the acyl transfer to happen, the hydroxyl of 1a must be much more nucleophilic than the hydroxyl of methanol, which is vastly more abundant as the solvent. The reaction happens spontaneously at room temperature over 24 h, converting ∼65% of 1a into the acetylated 3a, while a similar amount of 2a is converted into p-nitrophenol (PNP) (Figures S1 and S2).

1.

Structures of the reactants and products used in the acyl transfer model reactions.

Consistent with the importance of the ortho amino group in the enhanced nucleophilicity of the neighboring hydroxyl, neither 1b nor 1c is able to react with 2a under the same conditions (Figures S3 and S4). The lack of activity in 1b indicates that basicity of the amine is important to the activation of the neighboring hydroxyl. The lack of activity in 1c suggests that intermolecular general base catalysis (by the dimethylamino group meta to the hydroxymethyl) represents a negligible contributor to the activity of 1a. Similar to the previously reported serine protease models in the literature, 1a is completely unreactive toward 4-nitroacetanilide (2b), even at 80 °C (Figure S5). Thus, although 1a has sufficient nucleophilicity for the activated ester, it is inadequate for a less reactive aryl amide.

To place our catalytic dyad inside a substrate-tailored active site as in enzymes, we need to create a nanospace that contains not only the same catalytic groups but also a nearby space for binding the substrate. Furthermore, the nucleophilic hydroxyl needs to be positioned as closely to the carbonyl of the bound substrate (2a or 2b) as possible. Cram’s molecular model of the catalytic triad of chymotrypsin involves a heroic 30-step synthetic route. Thus, the construction of even an imperfect model of an enzyme active site could be enormously challenging, let alone the entire mimic of a functional enzyme.

Molecular imprinting is a technique to quickly build template-complementary imprinted binding sites in a polymer network, via covalent capture of template–functional monomer complexes through polymerization/cross-linking. − This method allows custom-designed molecular recognition to be accomplished in a straightforward manner via usage of appropriate template molecules and functional monomers that bind the templates through noncovalent and/or reversible covalent bonds. Molecularly imprinted polymers (MIPs) have many applications in chemistry and biology, − including in catalysis. −

To make the final catalyst resemble a globular protein, we performed molecular imprinting within surfactant micelles, affording water-soluble nanoparticles with a hydrophilic exterior and a hydrophobic interior (Scheme ). The three-step one-pot reaction involves surface-cross-linking of the template-containing micelles of surfactant 4 by the copper-catalyzed alkyne–azide cycloaddition with diazide 5 (step a), surface-functionalization of the resulting cross-linked micelles by another round of click reaction using monoazide 6 (step b), and free radical core-cross-linking of the micelles by divinylbenzene (DVB) initiated photochemically using 2,2-dimethoxy-2-phenylacetophenone or DMPA (step c). An attractive feature of micellar imprinting is the high fidelity of the imprinting process, with the imprinted micelles able to distinguish the addition, removal, and shift of a single methyl (or methylene) group during guest binding.

1. Preparation of a Synthetic Mimic of Serine Esterase/Protease through Molecular Imprinting of a Cross-Linked Micelle, with the Surface Ligands (i.e., Clicked 6) Omitted for Clarity .

^a The active site, not drawn to scale, schematically illustrates the key functionalities in the molecularly imprinted site.

Template 7 is color-coded according to the different purposes of its substructuresthe blue moiety is the space holder for the to-be-installed catalytic structure (8a–d) and the red for the to-be-bound substrate (2a,b). This molecule, due to its hydrophobicity, spontaneously enters the cationic micelles of 4. Cationic micelles are quite basic near their surface (pH ∼9.5 when the bulk solution is neutral) due to their electric potential. Thus, the template should enter the micelle in the anionic form, likely ion-paired with an ammonium headgroup of a neighboring surfactant. Its vinyl group allows it to be copolymerized with DVB and the methacrylate of 4 to become part of the cross-linked micelle core. The imine bond of polymerized 7 in nanoparticle NP1 is then hydrolyzed using 6 M HCl (step d in Scheme ), to afford an aldehyde group in the imprinted pocket of NP2. Reductive amination using 8a (steps e and f) installs the amine–alcohol dyad in the active site of NP3a, our serine esterase/protease mimic.

As discussed earlier, despite the enhanced nucleophilicity of the alcohol, 1a did not display catalytic turnovers. Remarkably, with the same amine–alcohol dyad in the molecularly imprinted active site, NP3a hydrolyzes 2a in enzyme-like catalysis (Figure a), turning over 970 of the substrate molecules in 90 min in a pH 7 buffer (Figure S9). Its Michaelis constant (K _m) is 67.3 μM and the catalytic turnover (k _cat) is 1.03 min^–1, affording a catalytic efficiency of k _cat/K _m = 255 M^–1 s^–1 (Table , entry 1).

2.

(a) Michaelis–Menten plot of the hydrolysis of 2a by NP3a in 25 mM HEPES buffer at 40 °C and pH 7.0. [NP3a] = 8 μM. (b) Hammett σ–ρ correlation in the hydrolysis of para-substituted-phenyl acetates catalyzed by NP3a. Reaction rates were measured in 25 mM HEPES buffer (pH 7) at 25 °C. [ester] = 50 μM. [catalyst] = 15 μM. σ values: p-NO₂, 0.78; p-CH₃CO, 0.52; p-Cl, 0.23; p-CHO, 0.22; p-H, 0.00; p-CH₃, −0.17.

1. Michaelis–Menten Parameters for NP3a at 40 °C.

entry	substrate	pH	kcat (min–1)	Km (μM)	kcat/Km (M–1 s–1)
1	2a	7.0	1.03 ± 0.01	67.3 ± 1.5	255
2	2b	7.4	0.0051 ± 0.0004	84.9 ± 23.3	1.0

^a[NP3a] = 8.0 μM in the Michaelis–Menten kinetic measurements (Figure S24).

There are at least three possible mechanisms for NP3a to catalyze the hydrolysis of 2a: (a) the amine-activated nucleophilic attack by the active site hydroxyl as hypothesized, (b) a direct nucleophilic attack by the amine followed by the acyl transfer to the adjacent hydroxyl, and (c) a general base-catalyzed nucleophilic attack of the carbonyl by an active site water molecule. One way to distinguish these mechanisms is through a Hammett plot. , For aryl ester hydrolysis in solution, the amount of negative charge developed on the phenol oxygen depends on the types of nucleophiles involved: a nucleophilic attack by a strong, anionic oxygen-based nucleophile (e.g., hydroxide or alkoxide) affords a reaction constant (ρ) of 1–1.2, by a neutral nitrogen a ρ of 2–3, and by a general base a ρ of 0.5–0.7. − The ρ value for the NP3a-catalyzed hydrolyses of substituted phenyl acetates is 1.52 (Figure b), more consistent with a nucleophilic attack by an anionic oxygen nucleophile, as in our proposed mechanism (Scheme , lower right nanoparticle). Essentially, the alcohol in NP3a behaves more like an alkoxide/hydroxide, similar to the active site serine in a serine protease (see below for additional discussion on the possible attack by hydroxide).

A general base-catalyzed ester hydrolysis due to breaking of the water O–H bond in the rate-determining step typically displays a primary solvent kinetic isotope effect (KIE) of k _H2O/k _D2O = 2–3. − Yet, the NP3a-catalyzed hydrolysis of 2a has a solvent isotope effect of 1.04 (Table S1), indicating that no breaking of the O–H bond occurs in the rate-limiting step, also consistent with our proposed mechanism.

It is clear at this point that moving the catalytic groups into a desolvated active site can boost their activities. Most amazingly, NP3a is found to hydrolyze aryl amide 2b catalytically near physiological conditions (Table , entry 2), which is completely impossible outside the active site. Both the amino group and the benzylic hydroxyl are critical to the catalysis, as no hydrolysis of 2b is observed with control nanoparticles NP3b–d prepared from a similar reductive amination of NP2 using 8b–d (Scheme ). Thus, not only is the microenvironment of the catalytic dyad crucial to the catalytic amide hydrolysis but also both the ortho amine and the neighboring hydroxyl remain critical.

Figures a and b show the pH profiles for the NP3a-catalyzed hydrolysis of 2a and 2b, respectively. For the hydrolysis of the activated ester, the reaction in buffer (i.e., background hydrolysis) and that in the presence of the nonimprinted nanoparticle (NINP, the control catalyst) are significantly slower than the catalyzed reaction (Figure a). For the hydrolysis of aryl amide 2b, the background hydrolysis and the hydrolysis by NINP are nondetectable and thus not shown in Figure b with log­(k) in the y-axis (see Figure S10 for the linearly plotted pH profiles).

3.

(a, b) pH profile for the hydrolysis of 2a (a) and 2b (b) under different conditions. NINP is the nonimprinted nanoparticle prepared without the template and is used as the control. All of the reactions were performed in HEPES buffer at 40 °C with [substrate] = 50 μM and [NP3a] = [NINP] = 10 μM. The hydrolytic yields for 2b were determined by LC-MS analysis. (c) Plausible mechanism for the hydrolysis of 2a and 2b in the active site of NP3a. Other intermediates may be involved as a result of proton transfer between the charged sites.

Catalytic hydrolysis of 2a involves first a transfer of the acetyl from the activated ester to the active site benzylic alcohol and then hydrolysis of the acylated catalyst (D), a nonactivated ester. Since the substrate has a particularly good leaving group (p-nitrophenolate), it is reasonable to expect that deacylation of the catalyst (steps 4–6) is the lower step(s) and rate-limiting. This picture is supported by the “burst” of p-nitrophenol release at the very beginning of the catalytic reaction (Figure S9). In addition, Figure a shows a monotonic increase in the activity of NP3a in the catalytic hydrolysis of 2a (Figure a, black triangle data points). Since the secondary rate constant for ester hydrolysis by hydroxide is 10^7–11 times faster than that by water, a higher solution pH is expected to lead to a faster hydrolysis of the acylated catalyst D.

Hydrolysis of 2b is understandably more difficult due to the much poorer leaving group. Its catalytic hydrolysis by NP3a exhibits a pH-dependent region (over pH 6–8) and a pH-independent region (over pH 8–10). Since the two substrates share the same acylated intermediate (D) but the hydrolysis of 2b is independent of solution pH over pH 8–10 while the hydrolysis of 2a is pH-dependent, the rate-limiting step in 2b is most likely acylation of the catalyst (steps 1–3). The pH profile supports such a picture. For the acylation, the active site amino group needs to be in the basic form (i.e., unprotonated) in order to activate the ortho hydroxyl for the nucleophilic attack of the carbonyl. The large change of activity over pH 6–7 in Figure b should correspond to the protonation of the active site amine. Once the amine becomes neutral (>pH 8), as long as the acylation of the catalyst is rate-limiting, the solution pH that affects the deacylation of the catalyst is irrelevant to the overall rate.

What could possibly make the same catalytic groups far more powerful inside the imprinted active site than in solution? Multiple factors might be involved. First, there is a strong (hydrophobic) driving force for the substrate to enter the active site of NP3a, evident from the 70–80 μM K _m values for the two substrates (Table ). Not only so, covalent imprinting has a high degree of fidelity in the imprinting process. , Given the structure of the template and its similar shape/dimension to the combined 8a and 2a or 2b, the nucleophilic hydroxyl in NP3a is anticipated to be positioned right next to the carbonyl of the bound substrate. Both factors should benefit the catalytic reaction greatly over the solution reaction that relies on random collision of the reactants to form the activated complex. Second, the hydrolyzed products are more hydrophilic than the substrate (2a or 2b). Differentials in their ability to occupy the active site would help the exit of the products and the entrance of the substrate, facilitating catalytic turnovers. Third, the intramolecular hydrogen bond between the ortho amino group and the nucleophilic benzyl alcohol should be significantly stronger in the hydrophobic active site of NP3a, as it does not need to compete with protic solvent molecules as in the solution. Additionally, the imprinted pocket is expected to template the two catalytic groups for the intramolecular hydrogen bond, again due to the similarity of the hydrogen-bonded 8a in comparison to the blue substructure of the template. A stronger intramolecular hydrogen bond would make the benzylic alcohol more nucleophilic than that in solution, evident from its activity toward aryl amide, observed only inside NP3a. Fourth, specific electrostatic interactions could be present in some of the intermediates (e.g., B and F in Figure c), and are expected to be much more effective in a hydrophobic microenvironment with a low dielectric constant than in a protic solvent. Indeed, specific electrostatic interactions between a reacting substrate and the functional groups in the enzyme active site have been considered a major basis for enzymatic efficiency.

The red substructure of the template (7) is the space holder for the (red-colored) substrate in the NP3a catalyst (Scheme ). In line with the shape selectivity, this serine protease mimic hydrolyzes aryl amides carrying various para- and ortho-substituents on the phenyl (Figure ). A small increase in the acyl chain is also tolerated, evident from the reasonable reactivity of para-nitrophenyl butyrate (2g). In contrast, the meta-substituted 2i and 2j display negligible activity toward NP3a, consistent with our design. Note that the dinitro-substituted 2j should be one of the most reactive amides in this group due to its double electron-withdrawing groups on the phenyl ring.

4.

Yields of aryl amide hydrolysis catalyzed by NP3a after 24 h in HEPES buffer (pH 7.4) at 40 °C. [Substrate] = 50 μM. [NP3a] = 10 μM. The yields were determined by LC-MS and HPLC analyses (Figures S11–S23).

In summary, moving the same amine–alcohol dyad from a solvent-exposed small molecule scaffold into a substrate-tailored imprinted hydrophobic pocket converts a catalytically incompetent enzyme model into a model enzyme mimicking serine esterase/protease. Neither the efficient catalytic turnovers for ester hydrolysis (Figure a) nor the ability to catalytically hydrolyze aryl amides (Figures and ) are present in the small molecule model (2a). These results suggest that true enzyme-like synthetic catalysts might not be very far from what chemists have accomplished already with small-molecule enzyme models. Effective implementation of catalytic strategies commonly used by enzymes, such as proximity, desolvation, reinforced hydrogen bonds, and effective electrostatic stabilization of charged transition states, could give chemists entry into the land of designable artificial enzymes. Although it is extremely challenging to build multiple catalytic features on a small-molecule organic scaffold through step-by-step organic synthesis, molecular imprinting enables facile construction of template-complementary binding pockets in a polymer network and, with proper template design and postfunctionalization, seems to be well-suited for the task.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.5c01302.

• Synthesis and characterization of materials, experimental details, LC-MS chromatographs, additional figures, and NMR and MS data (PDF)

The authors declare no competing financial interest.