Amyloid-reoriented enzyme catalysis

Abstract

Enzyme catalysis is essential for molecular transformations. Here, we make use of amyloid, a fibrillar aggregate formed by stacking peptides with β-sheet, which offers unique selectivity in enzymatic reactions. Azo-stilbene derivative (ASB), the amyloid-recognition motif, is incorporated into the substrate, which allows the amyloid consisting of Bz-Phe-Phe-Ala-Ala-Leu-Leu-NH₂ (BL7) to shield the substrates from the approaching enzyme. X-ray crystallographic analysis and structure-shielding effect relationship studies of BL7 reveal that the benzene rings present in the N-terminal benzoyl group and Phe1 side chain are particularly important for the shielding effect on the substrate. The finding results in a selective transformation system in which the reactive site close to ASB is protected by amyloid, while a site far from ASB is converted by the enzymes (trypsin, protein arginine deiminase [PAD], and Staphylococcus aureus V-8 Protease [Glu-C]). Further, the amyloid-shielded enzyme catalysis is compatible with an intact peptide, as the side chain of Tyr can be converted to the amyloid-recognizing motif. The enzymatic reactions combining amyloid provide unique selectivity for molecular transformation which may be used in diverse fields, including in synthetic chemistry.

Enzyme catalysis is essential for molecular transformations. Here, the authors show that incorporation of an amyloid recognition motif (azo-stilbene derivative) into the substrate allows the amyloid consisting of Bz-Phe-Phe-Ala-Ala-Leu-Leu-NH2 to shield the substrate from the approaching enzyme.

Introduction

Naturally-occurring enzymes efficiently, precisely, and sustainably catalyze chemical reactions in living organisms. Artificial manipulations, such as genetic engineering of the primary structure and combination with artificial small molecules, have markedly expanded the repertoire of applicable substrates and reaction patterns^1–3. These enzymatic reactions are currently indispensable in a variety of fields, including in synthetic chemistry. For example, many complex molecule syntheses^4–6, environmentally-friendly syntheses^7,8, and cost-effective industrial syntheses^9–11 have been achieved by combination with the enzymatic reactions. However, expanding the scope of enzyme catalysis remains an important challenge.

To date, such artificial interventions have focused on the enzyme. Here, we report a substrate “add-on” strategy to expand the scope of the enzyme reactions (Fig. 1a vs. b). Specifically, the structure as an add-on shields the substrate from the proximity of the enzyme. Amyloids, which have a rigid higher-order structure composed of β-sheet peptides¹², could be one such add-on structure. Here, we demonstrate that amyloid-shielded substrates are no longer under the control of the enzymes (Fig. 1b–1). In addition, we report the enzymatic reactions combining amyloid enabled regioselective conversion of the substrates, which was previously difficult to achieve (Fig. 1b−2).

Fig. 1. Conceptual illustrations to explain amyloid-reoriented enzyme catalysis.

a General enzyme catalysis. The substrate undergoes an enzymatic transformation to give a product. b This work. (1) Amyloid-shielded substrate is no longer under the control of the enzyme, and the structure is unchanged. (2) A potential modification site is shielded by the amyloid and is therefore inaccessible to the enzyme, while the other site far from amyloid undergoes enzymatic conversion. This results in a regioselective conversion.

Results

Identification of BL7 that shields substrate from PLE

We began by exploring the amyloids that shield the substrates from the enzymes. First, we selected an enzymatic reaction using pig liver esterase (PLE), a carboxylesterase (E.C.3.1.1.1) that hydrolyzes ester bonds¹³. As a substrate, we used an azo-stilbene (ASB) derivative (1, Fig. 2a) containing a phenyl acetate. ASB is a binder motif to the β-sheet characteristic of amyloid (cross-β-sheet)^14,15. We previously reported a catalysis system driven by amyloid-substrate complex exploited the binding of amyloids with ASB¹⁶. Ammonium ions attached to ASB were activated due to the close proximity to the amyloid catalyst formed by hexapeptide, Ac-Asn-Phe-Gly-Ala-Ile-Leu-NH₂ (NL6), which promoted the nucleophilic amine modifications in the acidic buffer, which is usually difficult to proceed. Here, we therefore investigated the effect of shielding 1 from PLE by forming the complex between ASB and NL6. As expected, the amyloid of NL6 showed a potent binding affinity for 1 possessing ASB (Fig. S1a). The ester bond of 1 was, however, hydrolyzed by PLE, regardless of the presence or absence of NL6, in which approximately 35% of 1 was consumed in both conditions (Fig. 2b and S2a, b). NL6 was suggested by this result to be inert in shielding the substrate from PLE. Inspired by the potential of phenyl groups to alter the amyloid strucuture¹⁷, we designed and synthesized Bz-Phe-Phe-Ala-Ala-Leu-Leu-NH₂ (BL7), which has two more benzene rings than NL6 (the benzoyl [Bz] group was introduced instead of the acetyl group at the N-terminus and Phe replaced Asn). The amyloid propensity of BL7 was verified by thioflavin-T (ThT) dye, the fluorescence intensity of which corresponds to the extent of cross-β-sheet of amyloid¹⁸, and transmission electron microscopy (TEM) (Fig. 2c). When BL7 was incubated in 5 mM HCl in DMSO-H₂O solution, the ThT fluorescence intensity was markedly strengthened compared with the control without BL7, and thick fibers of BL7 were clearly observed in the TEM analysis, supporting the amyloid-prone nature of BL7. In addition, BL7 amyloid had a potent binding affinity for 1 (Fig. S1b). Importantly, BL7 showed a shielding effect against the hydrolysis of 1 by PLE (Figs. 2b and S2c). Thus, compared with identical conditions but with the absence of amyloid (64%) or the presence of NL6 amyloid (66%), the recovery yield of 1 was increased to 94% in the presence of BL7 amyloid. We also analyzed the enzymological parameters of PLE for 1, and compared with a reference substrate, p-nitrophenyl acetate (pNPA), to which BL7 amyloid does not bind (Figs. 2d, e, and S1c). Michaelis–Menten plot for pNPA indicated that the values of Michaelis constant (Km) and Vmax of PLE in the presence of BL7 (100 µM) were 57.5 µM and 1.06 µM/s, respectively, comparable to those in the absence of BL7 (Km = 57.3 µM and Vmax = 0.93 µM/s) (Fig. 2d). The presence of BL7 is suggested by these results to have no effect on the enzymatic activity of PLE. In sharp contrast, when the substrate was 1, the results were significantly different depending on the presence or absence of BL7 (Fig. 2e). Reasonable Km and Vmax values of PLE were observed in the absence of BL7 (Km = 11.1 µM; Vmax = 0.296 µM/s), but the ester hydrolysis of 1 by PLE was significantly attenuated in the presence of BL7 (kcat/Km value: 1.14 µM^–1s^–1 vs. 8.60 µM^–1s^–1 in BL7-free condition). In addition, the degree of the attenuation of the reaction rate was in a concentration-dependent manner of BL7: the original reaction rate of PLE on 1 (10 µM) without BL7 was reduced to 87% with 5 µM BL7, 72% with 10 µM BL7, and 22% with 100 µM BL7 (Fig. 2f). BL7 amyloid is suggested by these results to shield substrate 1 from PLE through formation of the amyloid-substrate complex.

Fig. 2. Studies on PLE-catalyzed reactions.

a Hydrolysis of ester bond in substrate 1 (10 µM) possessing amyloid-binding ASB group. PB = phosphate buffer. b Recovery yields of 1 in the reaction conditions depicted in (a) without or with 100 µM of amyloid (NL6 or BL7). c Characterization of BL7 amyloid using ThT fluorescence assay (left) and TEM image (right). For ThT fluorescence assays, the mean ± SEM of relative fluorescence intensity values from three independent experiments (i.e., n = 3) are shown. d Michaelis–Menten analysis of PLE for pNPA without or with BL7 (100 µM). e Michaelis–Menten analysis of PLE for 1 without or with BL7 (100 µM). f Initial reaction rates of PLE for 1 (10 µM) with varied concentrations of BL7 (0–100 µM). g Substrate scope of PLE-catalyzed hydrolysis without or with BL7. All recovery yields were determined using analytical high-performance liquid chromatography (HPLC).

We investigated the substrate scope of the shielding effect from PLE by BL7 (Fig. 2g). In substrate 1a, in which an alkyl acetate was linked to ASB, the recovery yield of the substrate increased to 95% by the addition of BL7 under conditions where 1a was consumed by the PLE-mediated ester hydrolysis to a recovery yield of 35%. Similar results (BL7+ : 78%, BL7–: 6%) were obtained with substrate 1b in which the linkage structure to ASB in 1a was changed from the amide to an ether. Moreover, BL7 showed a sufficient shielding effect for glycine methyl ester (1c), which has ASB via the α-amino group (BL7+ : 85%, BL7–: 16%). Comparable results were observed with substrate 1 d, in which glycine of 1c was changed to leucine (BL7+ : 83%, BL7–: 7%). When the substitution was made with proline (1e), the hydrolysis of the ester by PLE itself became difficult to proceed, but the shielding effect by BL7 was detected (BL7+ : 100%, BL7–: 89%). A series of substrates possessing ASB were therefore appliable for the BL7-mediated shielding from PLE.

Shielding of peptidic substrates from enzymes

Next, we investigated the shielding effect of BL7 from enzymes on the peptidic substrates, composed of polyamide bonds (Fig. 3). As the enzyme, we adopted trypsin (E.C.3.4.21.4), which cleaves the amide bonds at the C-terminal side of arginine and lysine residues. We confirmed that BL7 amyloid is stable against trypsin (Fig. S3). The amyloid binding group (BGs) composed of ASB with a glycyl linker was introduced at the N-terminal side of arginine residue to give a pentapeptide, BGs-Arg-Ile-Ser-Val-Ala (2) (Fig. 3a). Michaelis–Menten analysis showed that the trypsin alone condition upon conversion of 2 to BGs-Arg (3) gave a canonical hyperbolic curve with Km and Vmax values of 10.9 µM and 67.2 µM/sec, respectively (Fig. 3b). The presence of trypsin inhibitor (TI) 4-aminobenzamidine¹⁹, significantly reduced the initial reaction rate (Vmax = 11.7 µM/sec), yet the plotting follows the Michaelis–Menten equation (Fig. 3c). In the presence of BL7, however, the Michaelis-Menten equation was not fitted, giving an S-shaped curve instead of a hyperbolic curve. The incompatibility with Michaelis–Menten plot agrees with the inhibitory mode of BL7, which acts on the substrate rather than on the enzyme. The extent of the reaction rate attenuation was dependent on the concentration of BL7, similar to that on PLE (Fig. 3d). Moreover, 2 was completely consumed by trypsin within 30 min through an exponential decay, whereas the recovery yield of 2 increased to 58% when 2 equivalents of BL7 were present relative to the substrate, and 2 was quantitatively recovered in the presence of 10 equivalents of BL7 (Fig. 3e). In the conditions with 10 equivalents of BL7, the recovery yield of 2 remained above 98%, even when the incubation time was extended to 24 h (Fig. S4). We also added 10 equivalents of BL7 to the solution with a reaction time of 5 min under the “No BL7” condition in Fig. 3e (more than half of 2 was degraded by trypsin), and monitored the recovery yield of 2 after the addition of BL7 (Fig. S5). As a result, the decrease in the recovery yield of 2 completely stopped after the addition of BL7 amyloid, indicating that BL7 forms a complex with 2 containing ASB within a short period of time. Furthermore, we investigated the effects of pH and ionic strength on the extent of substrate protection by BL7 against trypsin. Although it was to some extent affected by pH and ionic strength, BL7 maintained its strong substrate protection effect throughout (Table S2). On the other hand, to examine the substrate specificity of BL7, we performed trypsin-catalyzed reaction in the presence of BL7 in a mixed solution of 2 and a peptide in which the ASB of 2 was replaced by a Phe-Phe dipeptide (i.e., FFGRISVA) (Fig. S6). As a result, 2 was almost completely protected from trypsin by BL7, while most of peptide FFGRISVA was cleaved by trypsin to give FFGR without protection by BL7. This result supports that aromatic ring-rich motifs commonly present in peptides do not become amyloid recognition elements, and unintended shielding of non-target substrates is therefore minimal.

Fig. 3. Studies on trypsin-catalyzed reactions.

a Tryptic cleavage of amide bond at the C-terminal side of arginine residue in substrate 2 possessing the amyloid binding group (BGs) composed of ASB with a glycyl linker. The reaction was conducted in phosphate buffer (pH 8.0) at 30 °C. b Michaelis–Menten analysis of trypsin for 2. c Michaelis–Menten analysis of trypsin for 2 in the presence of TI (25 µM) or BL7 (100 µM). d Initial reaction rates of trypsin for 2 (10 µM) with varied concentrations of BL7. e Time-course of trypsin-catalyzed reactions of 2 (10 µM) in the absence (black line) or presence of BL7 (red line with filled squares: 20 µM, red line with white circles: 100 µM). f Substrate scope of trypsin-catalyzed reactions without or with BL7. All recovery yields were determined using analytical HPLC. *The product is BGs-K. g Comparison of the shielding effect of BL7 for 2 and 2 h. A_rel = (100 – recovery yield in the presence of BL7) / (100 – recovery yield in the absence of BL7).

We examined the substrate scope of the shielding effect by BL7 from trypsin (Fig. 3f). Various BGs-peptides (2a–2 g) were examined under conditions where the recovery yield of substrate 2 in the presence of trypsin was 2% without BL7 and 98% with BL7. Although the cleavage efficiency by trypsin varied depending on the substrate, there was universal observation of the shielding effect of BL7. Specifically, the shielding effect of BL7 was observed in overall BGs-peptides containing amino acids with polar functional groups such as Lys (2a, 2c, 2e, 2 g), Ser (2 d, 2e), Thr (2 g), Asn (2 d), Gln (2 g), His (2e), and Tyr (2 f). This result supports the idea that BL7 generally binds to BGs without interference from polar amino acids. The result that the shielding effect of BL7 is maintained not only on the substrates with C-terminal amides (2b, 2 d) but also with carboxylic acids (other than 2b and 2 d) also indicates that polar functional groups do not interfere with the binding of BL7 to BGs. Indeed, the binding affinity of 2e with many polar functional groups (Lys, Ser, His, C-terminal COOH) to BL7 was potent and comparable to that of 2 and 2a (Fig. S7). In addition, the shielding effect of BL7 was observed with substrate 2 g containing proline, which is often unfavorable for the binding with amyloid²⁰. The binding affinity of 2 g to BL7 was also comparable to that of 2 and 2a (Fig. S7). Moreover, a similar shielding effect by BL7 was observed when the trypsin cleavage site was substituted from Arg to Lys (2a vs. 2c). Furthermore, we examined the shielding effect of BL7 for substrate 2 h, in which the distance between the BGs and Arg residue in 2 was extended by one residue (Fig. 3g). To compare the shielding effect by BL7 between substrates, we defined a relative activity value (A_rel) that calculates the ratio of the cleavage reaction yield in the presence of BL7 to that in the absence of BL7. The A_rel value of 2 under the condition depicted in Fig. 3f was approximately 0.02. Despite the C-terminal amide bond of Arg of 2 h being less easily cleaved by trypsin compared with 2 in the absence of BL7 (24% recovery yield in 2 h vs. 2% in 2), the recovery yield of 2 h in the presence of BL7 was rather lower (76%) than that of 2 (98%). As a result, the A_rel value of 2 h was 0.31, significantly higher than that of 2. The result with A_rel suggests that the substrate shielding effect of BL7 was significantly reduced in 2 h due to the extended distance between BGs and Arg.

To verify the generality of the protection effect of BL7 on Arg-containing peptidic substrates, we subsequently used a PAD (EC 3.5.3.15), which catalyzes the citrullination of arginine residue, instead of trypsin (Fig. S8). As a result, BL7 amyloid shielded Arg near the BGs of the substrates from approaching PAD in high efficiency, similar to the case of trypsin. Furthermore, to examine the protection effect of BL7 on substrates containing reactive site other than Arg, we used substrates containing glutamic acid residues that have a side chain involved in the anionic functional group (instead of cationic side chain of Arg). Specifically, several substrates, in which an ASB-triazolylbutanoic acid linker (designated BGt) was linked near the N-terminus of Glu, were used (Fig. S9). As for the enzyme, we employed Glu-C (EC 3.4.21.19), which cleaves the amide bond at the C-terminal side of glutamic acid residue. As a result, BL7 amyloid potently protected the reactive sites near the ASB of the substrates from Glu-C. It was also confirmed that, as with trypsin, the substrate protection effect of BL7 was weakened by distancing the amyloid binding site (BGt) from the reactive site of Glu-C (Fig. S9b, substrates S3c vs. S3d). These results using PAD and Glu-C support the generalizability of the present method.

Mechanistic insights into the substrate shielding by amyloids

The three-dimensional structure of BL7 amyloid at the atomic resolution was determined by X-ray crystallographic analysis, which allowed us to conduct a docking study with substrate 2a (Fig. 4a, b). BL7 peptides were revealed to be stacked perpendicularly to the β-sheet axis via the intermolecular hydrogen bonds between the main chain amide bonds, forming the cross-β-sheet structure characteristic of amyloids (Fig. 4a). Notably, 2a was stably stuck in a cleft along the fibril axis (Fig. 4a, b; see Supplementary Movie 1 for molecular dynamics simulation). The cleft was formed by two BL7 peptides unit serving an N-terminal Bz group, the side chains of two Phe residues (Phe1 and Phe2), and the side chain of an Ala residue (Ala3) (Figs. 4b and S10). The side chain of Phe1 of one BL7 and the side chain of Phe2 of the other BL7 are arranged like a gate of the cleft. In addition, the benzene ring at the N-terminus of one BL7 hydrophobically interacts with the side chains of Ala3 and Leu5 of the other BL7 (Fig. S11). The ASB of 2a, bound in the cleft, was suggested to interact with the methylene moiety (i.e., CH–π interaction) of Phe2 side chain of BL7 (3.5 Å, Fig. 4b). The N-terminal Bz group and the side chains of Phe1, Phe2, and Ala3 of BL7 therefore seem to cooperatively play an important role in the shielding of the substrate.

Fig. 4. Insights into the shielding effect of BL7.

a, b The three-dimensional structure of BL7 amyloid determined by X-ray crystallographic analysis and the docking study with substrate 2a. a Side view. b Cross-sectional view. These structures are shown by sticks with oxygen, nitrogen, and carbon colored in red, blue, and light green (except for carbon in 2a, which is colored in gray). c Chemical modification studies of BL7 and the shielding effect of BL7 derivatives for 2a. A_rel = (100 – recovery yield in the presence of BL7 derivative)/(100 – recovery yield in the absence of BL7 derivative).

To verify the observation, we subsequently performed structure–shielding effect relationship studies of BL7 for 2a (Fig. 4c). All derivatives of BL7 (entries 2–9) formed the amyloid structures (Table S1). We initially investigated the contribution of each phenyl group present in Bz group, Phe1, and Phe2 for the shielding effect. We therefore examined the peptides in which each phenyl group was replaced with a methyl group (i.e., Bz-to-Ac, entry 2) and hydrogen atom (i.e., Phe-to-Ala, entries 3 and 4). The shielding effect was completely lost in the peptide in which the N-terminal Bz group was replaced with the acetyl group (A_rel = 1.1 vs. A_rel of BL7 = 0.061). This result supports the importance of the benzene ring at the center of the cleft consisting of two BL7 peptides (Fig. 4b). The shielding effect was also greatly weakened when Phe1 was replaced with Ala (entry 3, A_rel = 0.40), suggesting the importance of the benzene ring located at the gate of the cleft. The Phe2-to-Ala substitution, however, maintained the shielding effect to a large extent (entry 4, A_rel = 0.17). This is consistent with the docking result, which showed that the methylene moiety (rather than the benzene ring) of Phe2 interacts with ASB of the substrate. Subsequently, we investigated the contribution of Ala3, whose side chain interacts with the N-terminal benzene ring. When a hydrogen atom in the side chain of Ala3 was replaced with a hydroxy group (i.e., Ala-to-Ser substitution, entry 5), the shielding effect was significantly reduced (A_rel = 0.59). The hydrophobic interaction between Ala3 side chain and N-terminal benzene is thus suggested to be critical for the substrate shielding. In sharp contrast, the shielding effect was largely maintained (A_rel = 0.14) when the adjacent Ala4 was replaced with Ser (entry 6). This may be because, unlike Ala3, the side chain of Ala4 faces the outside of the cleft without the hydrophobic interaction with the N-terminal benzene ring. Moreover, substitutions of Leu5 and Leu6 with Ala did not result in a significant decrease in the shielding effect (A_rel = 0.12 and 0.14, respectively, in entries 7 and 8). Although the side chain of Leu5 faces the inside of the cleft, this result suggests that the hydrophobic interaction of the isobutyl group of Leu with the N-terminal benzene ring is largely tolerated by the methyl group (Fig. S11). The side chain of Leu6 faces the outside of the cleft, as does Ala4. Even when one Leu residue was deleted, the shielding effect was maintained to a considerable extent (A_rel = 0.17, entry 9).

Regioselective transformations of peptide substrates

Bifunctional molecules in which the binders for amyloid and enzyme of interest are covalently linked have been reported^21–25. The molecules can simultaneously bind to amyloid and the target enzyme due to the sufficient space between the two binders. Based on the reports, it is anticipated that a part of substrate close to BGs is shielded by amyloid, while the substrate portion far from BGs is approached by the enzyme (Fig. 5a). Indeed, the result that 2 h, in which an amino acid was inserted between BGs and Arg residue, showed a higher conversion yield of tryptic cleavage in the presence of BL7 than 2, in which BGs and Arg were directly linked, supports this idea (Fig. 3g). We therefore investigated the regioselective trypsin-catalyzed transformation of a peptide containing two arginine residues, BGs-Arg-Asn-Ala-Ile-Ser-Gly-Arg-Ala-Gly-Ile-Lys (4) (Fig. 5b). In the presence of trypsin without BL7, both amide bonds on the C-terminal side of the two Arg residues in 4 were cleaved. Namely, BGs-Arg-Asn-Ala-Ile-Ser-Gly-Arg (4a) was preferentially produced immediately after the reaction (t = 1 min), but at t = 5 min, the yield ratios of 4a and BGs-Arg (3) became comparable, and finally, after 15 min, 3 became the predominant product. In contrast, when trypsin and BL7 were present, product 4a, in which only the C-terminal amide bond of the Arg far from BGs was cleaved, was obtained as the final major product. When a peptide with Ala3 substituted with Ser (entry 5 in Fig. 4c) and a peptide lacking two benzene rings (NL6, Fig. 2b), which have low and no substrate protection effect, respectively, were used instead of BL7, the ratio of 3 increased because the amide bond on the C-terminus of the Arg near the BGs was cleaved as well as the C-terminal amide bond of the Arg far from the BGs (Fig. S12). The regioselective amide bond cleavage of the peptide is suggested by the result with BL7 to be possible by shielding the potential reaction site by amyloid. To further verify the regioselective conversion of 4, we applied PAD, which catalyzes the citrullination of arginine residue (Fig. 5c). In the absence of BL7, citrullination occurred almost simultaneously at the two Arg residues at t = 1 min, and almost all of the substrate was double citrullinated to give 4 d after 60 min. In the presence of BL7, the Arg residue distal to BGs underwent the citrullination, whereas the Arg residue adjacent to BGs remained unchanged, judging from the result that 4b was the major product at t = 60 min.

Fig. 5. Regioselective transformations of peptides enabled by the shielding effect of BL7 amyloid.

a Conceptual illustration of regioselective, enzymatic transformation with amyloid. The light blue part represents the amyloid binding group (BG). The substrate portion close to the BG is shielded by amyloid and is therefore unaffected by the enzyme (remaining light purple color), whereas the substrate portion far from the BG is converted by the enzyme (changing from light to dark purple color). b Trypsin-catalyzed reaction of 4. Reaction conditions: 4 (10 µM), BL7 (none or 100 µM), and trypsin in phosphate buffer (pH 8.0) at 30 °C. c PAD-catalyzed reactions of 4. Reaction conditions: 4 (10 µM), BL7 (none or 100 µM), and PAD in HEPES buffer (pH 7.6 containing 10 mM CaCl₂ and 5 mM dithiothreitol) at 37 °C. HPLC charts show the data at t = 60 min (detected at 475 nm). d Glu-C-catalyzed reactions of 5. Reaction conditions: 5 (10 µM), BL7 (none or 100 µM), and Glu-C in phosphate buffer (pH 8.0) at 37 °C. The HPLC (detected at 475 nm) and pie charts show the data at t = 24 h.

To demonstrate the versatility of the regioselective transformations of peptides using BL7, we performed Glu-C-catalyzed amide bond cleavage of substrate 5 containing two glutamic acid residues, Glu-Arg-Arg-Asn-Glu-Asn-Ser, which was linked to amyloid binding group (BGt) via the N-terminus (Fig. 5d). When Glu-C was present alone without BL7, the amide bonds at the C-terminus of Glu were non-selectively cleaved to produce BGt-Glu-Arg-Arg-Asn-Glu (5a) and BGt-Glu (5b) in 30% and 42%, respectively. Conversely, when BL7 was additionally present, the peptide 5a, in which the amide bond cleavage proceeded only at C-terminal side of Glu far from BGt, was given as the predominant product in 71% yield.

Tyrosine is a BG-handle

To enhance the utility of the amyloid-reoriented enzyme reactions, we further attempted to demonstrate the shielding effect of amyloid starting from an intact peptide without the pre-installation of the amyloid binding group (BG) (Fig. 6a). Bioconjugation methods, which modify a tyrosine residue present in the intact peptides and proteins, have been well studied²⁶. The reaction of diazonium salts with the Tyr side chain gives corresponding azo compounds^27,28. Thus, by reacting Ac-Tyr-Gly-Arg-Ile-Ser-Val-Ala (6) with 4-acetylbenzenediazonium hexafluorophosphate salt in a bicarbonate buffer (pH 9.0), the side chain of Tyr residue was converted to ASB structure to give 6a (HPLC chart labeled ‘reaction crude’). The amide bond at the C-terminal side of Arg residue in 6a was quantitatively cleaved by successive treatment with trypsin to give 6b (HPLC chart labeled ‘trypsin alone’). However, the coexistence of BL7 suppressed the conversion of 6a to 6b (HPLC chart labeled ‘trypsin + BL7’). This result suggests that BL7 bound to ASB constructed at the late stage at the site that was originally the side chain of Tyr residue, which shielded Arg residue of 6a from trypsin. Indeed, we confirmed that BL7 does not bind to the original substrate 6, but it does bind to 6a that has acquired the ASB motif (Fig. S13). The amyloid-reoriented enzyme reaction is therefore thought to be compatible with the intact peptide.

Fig. 6. Binding of BL7 to BG constructed at the late stage using an intact peptide.

a Reaction scheme (upper) and associated HPLC charts (lower, detected at 350 nm). Authentic 6a and 6b were synthesized by a stepwise solid phase peptide synthesis using Ac-aY-OH. b Application to substrate-selective amide bond cleavage.

Finally, we demonstrated the substrate-selective enzymatic reaction in combination with BL7 (Fig. 6b). When a mixed solution of 6 and 7 (a derivative in which Tyr in 6 was replaced with Phe) was treated with the diazonium salt, 6 was converted to 6a, while 7 remained intact. Successive addition of trypsin and BL7 to the resulting reaction mixture achieved selective cleavage of the C-terminal amide bond of Arg residue in 7 to give 7a, without cleaving the corresponding amide bond derived from 6. We also confirmed that BL7 did not bind to 7, which contrasts with the result that BL7 binds to 6a (Fig. S13). Substrate-selective amide bond cleavage was therefore shown to be possible.

Discussion

Realizing selectivity on the molecular transformations is a key issue in diverse fields, including in synthetic chemistry. We have leveraged amyloids, which are fibrillar aggregates formed by stacking peptides with β-sheets, creating unique selectivity. Structural motifs such as ASB have affinity for amyloids, and thereby when compounded with the motifs they can be regarded as amyloid-binding substrates. Amyloid consisting of NL6 (Ac-Asn-Phe-Gly-Ala-Ile-Leu-NH₂) was reported to promote the nucleophilic modifications of amine functionality present in the ASB-containing substrates under the acidic conditions, which is usually difficult to proceed due to the protonation of the amino group¹⁶. Thus, substrate amines were activated upon close proximity to the specific carbonyl oxygen of amide in NL6 amyloid, while amino groups of the compounds without binding affinity to NL6 amyloid remained inactive. Selective conversion of substrates possessing equivalently reactive amine functionalities was therefore possible in catalytic reactions using amyloids. Here, we report the amyloid-shielded selective transformation system that is complementary to the amyloid-mediated activation of amine. Amyloid of BL7, which has more benzene rings than NL6, deactivated the enzymatic reactions of ASB-containing substrates by shielding the reaction sites from the approaching enzyme. Interestingly, X-ray crystallographic analysis and structure-shielding effect relationship studies of BL7 have suggested that the benzene rings present at the N-terminal benzoyl group and the side chain of Phe1 (which are absent in NL6) are important for the shielding effect on substrates. Application of the amyloid-shielded system to the peptides enabled the regio- and substrate-selective transformations in which the reactive sites near the ASB of the substrate were protected by amyloid, while the sites remote from the ASB were converted by the enzymes.

The enzymatic reactions can be intervened upon not only by acting on the enzyme side, but also by acting on the substrate side. An enzyme inhibition strategy, in which a binder for the substrate was adopted instead of a conventional enzyme inhibitor (i.e., a binder for the enzyme), has been reported²⁹. The binder protected the target substrate while not inhibiting the enzymatic activity, which allowed selective inhibition on a substrate without interfering with the reaction of the enzyme with other substrates. In addition, the protection of the substrate from enzymes using molecularly imprinted nanoparticles has been demonstrated, which have nanospace to accommodate the substrate of interest^30,31. With molecularly imprinted nanoparticles that bind to unique peptide sequences of the substrates, the protection of one specific substrate as well as protection of a portion of long peptide substrate were achieved in the presence of the enzymes such as trypsin and protein kinases that naturally act on many substrates and sequences. These methods rely on tailor-made binding molecules that specifically recognize the structures of the target substrates. Moreover, there have been reports of host cages that encapsulate the frequently-occurring structures in the peptides as guest molecules. A cubic cage with internal Zn, which binds to an imidazole side chain of histidine, captured His-containing peptides and protected them from trypsin digestion³². Cucurbit[7]uril (Q7) has also been shown to capture the N-terminal Phe structure, and therefore selectively protected the peptides with Phe at the N-terminus from aminopeptidase N (APN)³³. Furthermore, pseudo-peptidic cages, which bind to EYE motif, protected the Tyr of the substrates from kinases³⁴. Our method employs ‘tagging’ (i.e., amyloid-recognizing ASB) for the shielding of the substrates, which differs from the previously-reported methods that recognize the unique structures of the substrates. The present method therefore allows the selection of substrates to be protected/not protected via tagging/not tagging. The availability of the option for such selections is advantageous for efficiently producing a variety of structures.

In summary, we introduced amyloid-reoriented enzyme catalysis. The amyloid-recognizing ASB was incorporated into the substrate, allowing the amyloid to shield the target portion of the substrate from the approaching enzyme. This resulted in a selective transformation system in which the reactive sites close to ASB were protected by amyloid, while the sites far from ASB were converted by the enzymes. In addition, the strategy of amyloid-shielded enzyme reaction was compatible with intact peptide, as the side chain of Tyr was convertible to the amyloid-recognizing motif. This method facilitates a unique selectivity on the molecular transformations with the enzymes in synthetic chemistry.

Methods

Preparation of stock solution of BL7 amyloid

DMSO solution of BL7 was diluted to half with 10 mM aqueous HCl (final concentration of BL7: 4 mM), and the mixed solutions were allowed to stand at room temperature for 30 min. The stock solution of BL7 amyloid was stored at −80 °C until use.

Trypsin-catalyzed reaction of 4

DMSO stock solution of 4 (TFA salt) at 1 mM (2.0 µL) and DMSO-HCl aq. stock solution of BL7 (0 mM or 4 mM, 5.0 µL) were added to 193 µL of 100 mM phosphate buffer at pH 8.0 (final concentration of 4: 10 µM, BL7: 0 µM or 100 µM), and the mixture was vortexed and left at room temperature for 5 min. Subsequently, 40-fold diluted stock solution of trypsin (2.0 µL) was added to the solution. After incubating the reaction mixture for a certain period of time at 30 °C, an aliquot of the solution (20 µL) was added to neat TFA (20 µL) to quench the reaction. The mixture (10 µL) was analyzed using analytical HPLC (detected at 475 nm). Yields were determined using the HPLC peak areas of 4, 4a, and 3 according to the following equation:

$$\mathrm{Yield}\left(\%\right)=\frac{\mathrm{peak}\mathrm{area}\mathrm{of}\mathbf{4},\mathbf{4a},\mathrm{or}\mathbf{3}}{\mathrm{sum}\mathrm{of}\mathrm{peak}\mathrm{areas}\mathrm{of}\mathbf{4},\mathbf{4a},\mathrm{and}\mathbf{3}}\times 100$$

Trypsin-catalyzed reaction starting from 6

DMSO stock solutions of Ac-YGRISVA (6) (20 mM, 5.0 µL) and 4-acetylbenzenediazonium hexafluorophosphate (100 mM, 1.0 µL) were added to 94 µL of bicarbonate buffer (0.1 M, pH 9.0), and the mixture was left at room temperature for 30 min. The resulting reaction mixture was then 100-fold diluted with phosphate buffer at pH 8.0. DMSO-HCl aq. stock solution of BL7 (0 mM or 4 mM, 2.5 µL) and 5-fold diluted stock solution of trypsin (2.0 µL) were added to 95.5 µL of the solution. After incubating the reaction mixture for 1 h at 30 °C, an aliquot of the solution (20 µL) was analyzed using analytical HPLC (detected at 350 nm).

X-Ray crystal structure determination of BL7

Needle-like crystal of BL7 was obtained from formic acid solution by slow evaporation. Crystal was cryoprotected by soaking in Santovac® 5 Cryo Oil (Hampton Research, Aliso Viejo, CA) and being flash frozen in liquid nitrogen. X-ray diffraction data sets were collected at a wavelength and temperature of 1.00 Å and 100 K, respectively, using synchrotron radiation on the BL26B1 beamline (EIGER X 4 M detector, DECTRIS, Philadelphia, PA) by SPring-8, in Hyogo Japan. The initial structure was obtained by molecular replacement with Phaser using the polyalanine model (AAAAAA) of FGTGFG segment from the nucleoporin p54 (PDB ID: 7N8R) as a search model. The structure was refined with REFMAC5 and the quality of the final structure was assessed with MolProbity (Ramachandran statistics are as follows: Favored 100%, Allowed 0%, Outliers 0%).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

T.S. and Y.S. designed research; T.S., F.M., K.Y., and D.S. performed research; T.S., F.M., K.Y., D.S., and Y.S. analyzed data; and T.S. and Y.S. wrote the paper.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The atomic coordinate and structure factor have been deposited in the Protein Data Bank under the accession code 9JBL. All other data, including experimental procedures and HPLC charts, are provided in the Article, Supplementary Information and from corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-58536-5.