Info

Design of Enzymes

9.2.1.1 Grafting Catalytic Sites in Proteins

A catalytic site with a triose phosphate isomerase activity was grafted onto a ri-bose binding protein [1]. Computational design was used to create a substrate binding site and to position catalytic residues at the active site. Finally, a powerful in vivo selection in E. coli allowed the isolation of active variants with rate enhancements of 105 to 106 when compared to the uncatalyzed reaction.

Similarly, by combining in silico modeling of protein insertions and deletions with in vivo selection of variants for antibiotic resistance allowed a metallohydro-lase to be converted into a cefotaxime hydrolyzing enzyme [2].

9.2.1.2 Endowing Enzymes with Two Catalytic Activities in a Single Domain

Endowing an enzyme with two catalytic activities was achieved using sequence-catalytic activity relationships or using sequence-structure-catalytic activity relationships [3]. Recent experiments tend to show that the simplest strategy to enlarge the spectrum of catalytic activities of a polypeptide may be to direct its evolution towards a new catalytic activity (Fig. 9.2).

Directed evolution of a library of more than 107 Thermus aquaticus DNA-polymerase I mutants by in vitro selection for RNA-dependent DNA-polymerase activity (Fig. 9.3) yielded the isolation of DNA polymerase variants that copy both DNA and RNA with similar catalytic efficiencies [4]. This result is astonishing as the experimental evolution procedure did not include a selection for DNA-dependent DNA-polymerization.

In another study, 2 x 105 T. aquaticus DNA polymerase I mutants within the highly conserved motif A were selected in vivo for the wild-type activity, DNA-dependent DNA-polymerization. Screening of selected clones for DNA-dependent RNA-polymerization activity allowed the identification of variants with a 103-fold greater ribonucleotide incorporation efficiency than that of the wild-type enzyme. Ribonucleotide incorporation was still up to ten-fold less efficient than deoxyribo-nucleotide incorporation [5]. Direct selection for RNA-polymerization using phage display allowed the isolation of a variant of the Stoffel fragment of T. aquaticus DNA polymerase I with similar catalytic efficiencies for RNA- and DNA-polymerization on DNA templates. These catalytic efficiencies were found to be

Fig. 9.3 In vitro selection for catalytically active proteins using phage display The in vitro selection for enzymes includes a selection for foldability. An unfolded polypeptide (yellow), a folded polypeptide with a low catalytic activity (green) and a folded polypeptide with a high catalytic turnover (blue). Substrates are coupled to purified phage particles. The in vitro selection for catalysis relies on isolation by affinity chromatography for the product bound to phage-enzymes. If the polypeptide is unfolded or inactive, the substrates will not be converted to products; no particles will be

9.2 Design of Proteins from Natural Scaffolds | 271

Fig. 9.3 In vitro selection for catalytically active proteins using phage display The in vitro selection for enzymes includes a selection for foldability. An unfolded polypeptide (yellow), a folded polypeptide with a low catalytic activity (green) and a folded polypeptide with a high catalytic turnover (blue). Substrates are coupled to purified phage particles. The in vitro selection for catalysis relies on isolation by affinity chromatography for the product bound to phage-enzymes. If the polypeptide is unfolded or inactive, the substrates will not be converted to products; no particles will be purified by affinity chromatography. If the polypeptide has a low catalytic activity, the phage particle will not be isolated efficiently by affinity chromatography for the product. If the polypeptide has a high catalytic turn-over, multiple substrates are converted into products. Because of the chelate effect, this phage particle derivatized by multiple products will be efficiently isolated by affinity chromatography. The genes encoding active enzymes are located within the phage particles. This strategy can be applied to libraries of more than 107 variant enzymes.

less than ten-fold lower than that of DNA-dependent DNA-polymerization of the wild-type Stoffel fragment [6].

9.2.1.3 Grafting Allosteric Sites to Regulate Enzyme Activity

Regulation of an enzyme's activity was achieved by introducing mutations that allowed the binding of chosen ligands which could act as regulators.

Libraries of beta-lactamase variants with mutations in loops surrounding the catalytic site were constructed and allowed the isolation of lactamases binding various proteins such as streptavidin, ferritin, beta-galactosidase or antibody fragments [7]. This approach has been used as a method to titrate an antigen (the prostate-specific antigen, PSA) in solution by measuring the lactamase activity. In fact, the enzyme was engineered so as to bind an anti-PSA antibody fragment, forming a complex which was found to inhibit the enzymatic activity by 90% [8].

In another report, beta-lactamase was inserted randomly within a maltose-binding protein. The fusion proteins were selected for ampicillin resistance. Further screening for maltose-dependent activity allowed the isolation of lactamase variants with a 600-fold improved rate in the presence of maltose [9].

272 | 9 Protein Design 9.2.2

Design of Binding Proteins

The key points in designing binding proteins are (i) to choose a starting protein that will be used as a scaffold on the basis of its favorable biophysical properties and (ii) to define a mutagenesis scheme that will allow generation of the new binding property while minimizing negative effects on the foldability of the protein. It has been known for a long time that proteins can tolerate mutations, especially if mutated amino acids are located at their surface [10].

Antibodies, in particular their variable domains, are among the most studied proteins as they exemplify how a protein can show extraordinary adaptability to bind very different kinds of ligands, such as proteins, peptides, nucleic acids, sugars, etc. In antibodies, the scaffold is constituted of b-strands that are connected by six loops. These loops are highly variable in length and in sequence from antibody to antibody and provide the diversity needed for binding the ligand, while the immunoglobulin fold remains constant. Many attempts have been made to recreate this concept in vitro with antibodies. A study involving randomization of the most significant loop for antigen-binding and diversity (CDR-H3) [11] opened the door to fully synthetic libraries of antibodies. In this work, the authors showed that the randomization of one loop followed by selections by phage display was sufficient to convert an antibody specific for the human tetanus toxicoid antigen into antibodies specific for fluorescein. This approach was then extended to several loops [12, 13, 14]. With these libraries it was possible to isolate new antibodies with desired specificities for various ligands, often with dissociation constants in the nano- or even picomolar range.

Unfortunately, antibodies or their derivative fragments are often difficult to produce because of their molecular complexity and/or their stability which do not make them ideal molecules for nontherapeutic applications. For these reasons, alternative methods have been developed using other natural proteins engineered via a combinatorial mutation/selection approach. The aim was to challenge antibodies for their favorable properties while removing their disadvantages. Many projects have been undertaken during the last decade in this field and we will only focus on those where a three dimensional structure is available for binders generated from a designed scaffold alternative to antibodies.

The staphylococcal protein A is a bacterial receptor that binds the Fc region of immunoglobulin G. It has repetitive subunits of about 58 amino acids that are individually folded into a three-helix bundle (Fig. 9.4A), that is highly soluble and stable. Isolated domains, such as the Z-domain, were shown to tolerate multiple amino acids substitutions at the binding area involved in Fc recognition [15]. Combinatorial libraries corresponding to the randomization of thirteen residues, located into two a-helices of the Z-domain at the binding surface, were used for selections against three different proteins [16]. This work showed that the binding specificity of the Z-domain was changed from Fc to Taq polymerase, insulin or apolipoprotein recognition. Circular dichroism experiments indicated that these engineered binders (called affibodies) had a secondary structure similar to

Fig. 9.4 (A) Superimposition of structures of a complex Zspa-1 affibody-ligand (PDB ID: 1LP1) (blue) and ofwild-type protein Z domain (PDB ID: 1Q2N) (green); ligand (yellow). (B) Superimposition ofstructures of a complex FluA anticalin-ligand (PDB ID: 1N0S) (blue) and of wild-type bilin binding protein (PDB ID: 1BBP) (green); ligand (yellow) (fluorescein). (C) Superimposition of structures of a complex neocarzinostatin-

Fig. 9.4 (A) Superimposition of structures of a complex Zspa-1 affibody-ligand (PDB ID: 1LP1) (blue) and ofwild-type protein Z domain (PDB ID: 1Q2N) (green); ligand (yellow). (B) Superimposition ofstructures of a complex FluA anticalin-ligand (PDB ID: 1N0S) (blue) and of wild-type bilin binding protein (PDB ID: 1BBP) (green); ligand (yellow) (fluorescein). (C) Superimposition of structures of a complex neocarzinostatin-

ligand (3.24 binder) (PDB ID: 2CBO) (blue) and of wild-type neocarzinostatin (PDB ID: 1NCO) (green); ligand (yellow) (testosterone hemisuccinate). (D) Superimposition of structures of a complex designed ankyrin-ligand (PDB ID: 1SVX) (off7 binder) (blue) and of a wild-type ankyrin (PDB ID: 1AP7) (green); ligand (yellow) (maltose binding protein).

the native Z-domain suggesting that their fold is similar. Initial affinities obtained were in the micromolar range. With an affinity maturation step, via random substitution of six residues and a new selection, it was possible to obtain binders with dissociation constants around 40 nM, against the Taq polymerase [17] which is similar in strength to the association of the parent staphylococcal protein A with the Fc fragment. Binders with affinities in the nanomolar range were also obtained against several other protein targets such as the extra cellular domain of the HER2 receptor [18] with dissociation constants as low as 22 pM [19]. The concept of re-directing the specificity of a natural protein was thus clearly demonstrated in this case. However, the foldability of the resulting binders was not demonstrated. To answer these questions structural studies were undertaken for several affibodies, either alone or in complex with their respective partner, by

NMR [20], [21] and by X-ray crystallography [22]. The structures for the binder Zspa-i revealed that its interaction surface has some characteristics comparable to those observed in antibody-protein complexes (size of the surface area and hydrogen bonding sites) [22] and ten of the thirteen randomized residues were indeed interacting with the target. However, these structural data [20] and biophysical characterizations [23] showed that uncomplexed affibodies behave like a molten globule and that folding occurs upon binding to the target protein. By contrast, it was shown that other affibodies (ZTaq and anti-ZTaq) were well structured, folded even uncomplexed [21] and had higher melting temperature than for ZSPA-1 [24]. These studies showed that individual structures of affibodies are superimposable, indicating that the scaffold of the wild-type Z-domain was preserved after mutagenesis, at least in the complex. However, for the same scaffold, the assumption that foldability is retained after surface substitutions have been made, was verified in one case and not in the other. This suggests that selection for a binding property does not mean selection for foldability, at least for the isolated binders.

The bilin-binding protein, a member of the lipocalin protein family has been recruited as a scaffold for the development of new affinity reagents called ''antica-lins''. This protein of 174 amino acid residues folds into a b barrel with eight strands connected by loops which are involved in the binding of biliverdin (Fig. 9.4B). These loops indeed form a rather wide and shallow pocket for the hapten. Libraries of lipocalin mutants were created by randomization of sixteen residues distributed across four loops and were used for selection against several low molecular weight compounds. Binders were obtained by phage display with dissociation constants of 35 nM and 295 nM for fluorescein [25] and digoxigenin [26], respectively. The affinity was further improved with a resulting dissociation constant of 12 nM for digoxigenin after an affinity maturation step [27] and 1 nM for fluorescein by rational design [28]. Biophysical characterization of one mutant by circular dichroism showed that some of the residues in a b-strand and a-helical conformation were the same as those deduced from the crystallographic structure of the parent protein. A structural study of the mutant FluA in complex with flu-orescein further supported the idea that lipocalin can be mutated in the targeted loops without significantly disturbing its overall conformation [29]. Indeed, the major perturbations were observed in the loops. Hence, the foldability of the resulting variants was similar to the wild-type lipocalin.

Neocarzinostatin, an enediyne-binding chromoprotein of 113 amino acids consists of seven b-strands forming a b-sandwich, an architecture similar to antibodies (Fig. 9.4C). However, the binding of the chromophore, a small organic compound, involves thirteen amino-acids that constitute a binding pocket. To explore the potential of this pocket to bind other haptens, several libraries corresponding to randomization of up to thirteen residues were created and used for selection by phage display [30]. Variants able to bind specifically testosterone have been isolated. These binders showed nanomolar dissociation constants for strep-tavidin bound testosterone and micromolar KD values for the free soluble form of testosterone. Thus, it is possible for this scaffold to change its recognition proper ties to bind an unrelated ligand. The three dimensional structures of several mutants were obtained to understand the structural basis of the testosterone recognition [31]. These studies revealed that indeed the structure for the binder 3.24 was retained in the free and complexed forms. Uncomplexed binders 1a.15 and 4.1 were found to have the same structure as the wild-type protein, while they formed dimers when complexed with the ligand. Hence, the foldability of the neocarzi-nostatin was also retained after massive randomization of its amino acids.

Some proteins such as ankyrins are composed of several repeated motifs. A modular approach to library construction has been performed using these kinds of motifs. In ankyrins the repetitive structural units consist of twenty to forty amino acids folded in a b-turn and two a-helices. In contrast to other described scaffolds, for which a strictly natural protein was used as a starting point for library construction, the authors first designed self-compatible repeat modules [32], [33] by consensus analysis of naturally existing ankyrin repeats. Biophysical studies showed that this strategy produced well expressed, soluble, stable and folded proteins [34]. The three dimensional structure of a consensus ankyrin, with three central repeats and two N- and C-terminal caps, showed that the designed ankyrin retained the fold of natural ankyrins [35]. Hence, using the concensus sequence libraries were constructed with two or three randomized ankyrin repeat domains comprising six diversified residues [33]. These libraries were used for selections by ribosome display. Binders with nanomolar dissociation constants were obtained for maltose binding protein, two eukaryotic protein kinases [36] and a bacterial kinase [37]. The three dimensional structures of two binders, in complex either with maltose binding protein [36] (Fig. 9.4D) or aminoglycoside phosphotransferase [38], were obtained. These structures further confirmed that foldability was retained for the binders compared to wild-type ankyrins.

Diabetes Sustenance

Diabetes Sustenance

Get All The Support And Guidance You Need To Be A Success At Dealing With Diabetes The Healthy Way. This Book Is One Of The Most Valuable Resources In The World When It Comes To Learning How Nutritional Supplements Can Control Sugar Levels.

Get My Free Ebook


Post a comment