Letters Cite This: ACS Chem. Biol. XXXX, XXX, XXX−XXX
Monobody-Mediated Alteration of Lipase Substrate Specificity Shun-ichi Tanaka,†,‡,§,∥ Tetsuya Takahashi,‡ Akiko Koide,†,⊥,# Riki Iwamoto,∥ Satoshi Koikeda,‡ and Shohei Koide*,†,⊥,∇ †
Department of Biochemistry and Molecular Biology, The University of Chicago, Chicago, Illinois 60637, United States Frontier Research Department, Gifu R&D Center, Amano Enzyme, Inc., Gifu 509-0109, Japan § Ritsumeikan Global Innovation Research Organization, Ritsumeikan University, 1-1-1 Noji-higashi, Shiga 525-8577, Japan ∥ Department of Biotechnology, College of Life Sciences, Ritsumeikan University, 1-1-1 Noji-higashi, Shiga 525-8577, Japan ⊥ Perlmutter Cancer Center, New York University Langone Medical Center, New York, New York 10016, United States # Department of Medicine, New York University School of Medicine, New York, New York 10016, United States ∇ Department of Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, New York 10016, United States ‡
S Supporting Information *
ABSTRACT: Controlling the catalytic properties of enzymes remain an important challenge in chemistry and biotechnology. We have recently established a strategy for altering enzyme specificity in which the addition of proxy monobodies, synthetic binding proteins, modulates the specificity of an otherwise unmodified enzyme. Here, in order to examine its broader applicability, we employed the strategy on Candida rugosa lipase 1 (CRL1), an enzyme with a tunnel-like substrate binding site. We successfully identified proxy monobodies that restricted the substrate specificity of CRL1 toward short-chain fatty acids. The successes with this enzyme system and a β-galactosidase used in the previous work suggest that our strategy can be applied to diverse enzymes with distinct architectures of substrate binding sites.
E
and monobodies, synthetic binding proteins based on the 10th human fibronectin type III (FN3) domain, we successfully altered the enzyme’s substrate specificity for its transgalactosylation reaction and selectively enhanced the production of short oligosaccharides. The project was designed based on a general, but not atomic-level, understanding of the mechanism of substrate recognition by β-galactosidases in which discrete subsites recognize different sugar moieties within a substrate. By generating a monobody that occupies an appropriate subsite, thereby blocking the binding of oligosaccharides that requires the monobody-occupied subsite, we restrict the range of substrates upon which an enzyme acts.6 Because our strategy does not modify the targeted enzyme itself, it can be applied to an enzyme system that does not have either a heterologous expression system or detailed mechanistic knowledge. Furthermore, monobodies with desired targetbinding profiles can be readily identified through a highthroughput protein design platform,7 eliminating the need for a high-throughput enzyme assay. Therefore, our strategy should be an attractive alternative to conventional methods for
nzymes accelerate chemical reactions under mild, aqueous conditions, and thus their investigation is an important branch in chemistry and biology. Enzymes are widely used in industrial applications, including food, pharmaceuticals, dairy, and textiles. 1,2 Naturally occurring enzymes often are suboptimal for a specific application of interest, and, hence, their properties must be altered. However, engineering an enzyme toward desired catalytic properties remains a challenging task. Advances in recombinant DNA techniques allow us to create a polypeptide with any amino acid sequence; yet, our current knowledge of the sequence−structure− function relationships of enzymes is still at its early stages for de novo and/or rational enzyme design.3 Directed evolution is a powerful tool for engineering enzymes without the use of detailed mechanistic knowledge and has led to many successful examples.4 However, directed evolution experiments are often hampered if an efficient production system in a suitable heterologous host (e.g., Escherichia coli) and a high-throughput enzyme assay are unavailable.5 Thus, there remain many enzyme systems for which conventional enzyme-engineering approaches are ineffective or impractical. We have recently established a new strategy for altering enzyme specificity with proxy binding proteins directed to the substrate-binding site.6 Using an example of a β-galactosidase © XXXX American Chemical Society
Received: April 24, 2018 Accepted: May 14, 2018 Published: May 14, 2018 A
DOI: 10.1021/acschembio.8b00384 ACS Chem. Biol. XXXX, XXX, XXX−XXX
Letters
ACS Chemical Biology
the CRL1 substrate-binding site does not have discrete subsites for discriminating hydrocarbon chains of different lengths. Taken together, although chain length specificity of CRL1 and its homologues has been altered by structure-guided mutations of residues lining the substrate-binding tunnel,16−18 it is unclear whether a proxy monobody can achieve a similar effect. Thus, we considered alteration of the CRL1 substrate specificity an interesting and informative challenge for our enzyme engineering strategy. We generated a series of monobodies directed to CRL1 by performing combinatorial library selection using phage- and yeast-display technologies by following previously established procedures.7 The closed form of CRL1, which is monomeric,15 was isolated with size exclusion chromatography, and then used in the presence of Triton X-100 throughout monobody generation with the hope of enriching monobodies that are bound to the catalytically active state of CRL1. The initial unbiased selection yielded a total of 11 unique monobodies (see Figure S1a in the Supporting Information). Despite the sequence diversity in these monobodies, all of them are bound to an overlapping surface (“epitope”) of CRL1, as determined in competitive binding experiments using purified monobody samples (data not shown). These monobodies showed no detectable effects on CRL1 catalytic activity (data for only Mb(CRL1_S05) is shown in Figures 2c and 2d; for brevity, hereafter, in the text, we will use abbreviated names for monobodies where the “CRL_” segment is omitted), suggesting that they are “inert” monobodies that are bound to a location that is distant from a catalytically important site but somehow strongly attracted monobodies. To obtain monobodies that affect CRL1 catalysis, we performed the second selection campaign in which we used Mb(S05) to mask the dominant epitope. This selection yielded 11 additional monobodies (Figure S1a), and of them we identified three different types of monobodies, in terms of their effect on CRL1 activity. Mb(S16) potently inhibited CRL1 activity toward all chain lengths by reducing VMAX, as tested using synthetic p-nitrophenyl (pNP)-esters with a single aliphatic chain as substrates (see Figure 1, right panel and Figure 2c). We used these pNP esters for quickly characterizing the chain-length specificity at the scissile fatty acid binding site, because assays using triacylglycerides are labor-intensive and low throughput (see the Methods section). We observed inherently low activity of CRL1 toward the C6 substrates, as previously reported.8 Note that panels in Figures 2c and 2d use different horizontal scales to clearly show effects of monobodies. Mb(L18) showed a desired profile of a specificity modifier, as it significantly reduced the activity of CRL1 toward the chain lengths of C10 and longer, whereas it had minimal effects on C8 and shorter (Figure 2c). The inhibitory effect was primarily due to reductions of VMAX, whereas it marginally affected KM. The binding of Mb(L18) to CRL1 was competed by Mb(S16), suggesting that their epitopes are overlapped, although their functions on CRL1 activity are distinctly different (see Figures 2b and 2c). Mb(S19) did not show significant effect on CRL1 catalytic activity and its binding was not competed with either Mb(S16) or Mb(L18) (Figures 2b and 2c). Further selection campaign to obtain monobodies that are competed by Mb(L18) but have different profiles produced two additional inhibitory monobodies, Mb(L23) and Mb(L24) with a single amino acid difference between them (see Figure S1a, as well as Figure 2c). In total, we generated 24 monobodies with three
engineering substrate specificity, particularly for restricting substrates to smaller species among natural substrates. However, to date, we have shown its effectiveness with only a single enzyme system. Here, we tested the strategy on restricting the substrate specificity of a lipase, lipase 1 from Candida rugosa (CRL1),8 which is an enzyme that is distinctly different from the βgalactosidase used in our original work. CRL1 belongs to a family of triacylglycerol lipase (EC 3.1.1.3) and catalyzes both hydrolysis and synthesis of triacylglycerides and several other esters. It is highly active toward broad substrates, in terms of substrate size (short to long fatty acids) and in terms of substrate shape and chemistry (saturated and polyunsaturated fatty acids),9 and it is among the most widely used enzymes in industrial applications, such as food processing, pharmaceutical synthesis, oil refining, flavor development, and oil wastewater treatment.10,11 However, the broad specificity is not desirable for the purpose of the production of a small number of defined products. In fatty acid production, short-chain selectivity is often desired to produce ingredients in flavors and fragrances.12 For example, in ripening and enhancing cheese flavor, only short-chain fatty acids that give rise to cheese flavor should ideally be produced from a mixture of lipid substrates. In reality, however, like most lipases, CRL1 also hydrolyzes long-chain triglycerides and produce long-chain fatty acids that contribute to soapy flavor. Thus, tailoring the substrate specificity of CRL1 toward shorter triglycerides chains is an important goal. Although CRL1 and BgaD-D, which is the β-galactosidase used in our previous study, both catalyze hydrolysis of their respective substrates and also conjugation reactions (transesterification and transgalactosylation, respectively), these two enzymes differ substantially in terms of the architecture of substrate binding site. The substrate-binding site of BgaD-D consists of shallow, surface-exposed pockets readily accessible by proxy monobodies,13 whereas the scissile substrate-binding site of CRL1 is a deeply buried tunnel whose inner surface is unlikely to be directly accessible by monobodies (see Figure 1).14 Like most lipases, CRL1 has a mobile segment (“lid”) that
Figure 1. Schematics showing a working model for substrate recognition by CRL1, developed from crystal structures of CRL1 (PDB IDs: 1LPP and 1TRH).14 CRL1 has three hydrophobic sites for substrate recognition: one with deep tunnel-like architecture for the scissile fatty acid (FA) and two others as hydrophobic patches for the nonscissile FAs. The enzyme cleaves the ester linkage for the scissile FA, marked with the red triangle. The mobile segment “lid” is shown in pink.
covers the active site and keeps the enzyme in the closed and inactive form (Figure 1, left panel). The conversion from the closed form to the open form is slow in aqueous solution, whereas the exposure to a solute containing a hydrophobic moiety such as a detergent can induce lid opening, make the active site accessible to the substrate, thus activating the lipase (Figure 1, middle and right panels).15 Furthermore, the level of discrimination needed for CRL1 (one or two hydrocarbon units) is much finer than that for BgaD-D (a sugar unit), and B
DOI: 10.1021/acschembio.8b00384 ACS Chem. Biol. XXXX, XXX, XXX−XXX
Letters
ACS Chemical Biology
Figure 2. Monobodies binding to CRL1. (a) KD values of representative monobodies determined using yeast surface display titration in the absence (no Triton) or presence (+ Triton) of detergent. The titration data are shown in Figure S2 in the Supporting Information. (b) A matrix plot summary of competition between different monobodies in CRL1 binding. Binding of monobodies indicated at the top to CRL1 in the presence of a competitor monobody (left) is shown. The gray boxes indicate significant inhibition (
