Laboratory Home | Research | Publications | Personnel | Links | Gallery

Research

The research interests in the lab encompass the design, synthesis, and applications of agonists, antagonists and artificial biocatalysts from a chemist’s perspective. Some of the projects in progress in the lab are summarized here.

Drafting an Enzyme to Selectively Assemble its Best Inhibitor from an Array of Building Blocks.

Generation of lead compounds is central to the drug discovery process. In collaboration with the Sharpless (TSRI), Finn (TSRI) and Taylor (UCSD) groups, we have developed an approach that engages the protein target in the actual selection of building blocks from a given array to directly synthesize its own inhibitor. This approach relies on the enforced proximity that results from binding two adjacent ligands decorated with complementary reactive groups in the binding cleft of the casting enzyme, promoting a fusion reaction between the two building blocks (which would otherwise be slow in the absence of the template). In principle, the enzyme could either catalyze the formation of a privileged product with multiple turnovers or be limited by product inhibition. In the latter case, the higher affinity products would then serve as lead compounds for drug discovery.
The reactive building blocks have azide or acetylene groups that combine efficiently, when in close proximity, to form triazoles via a 1,3-dipolar cycloaddition reaction (Fig. 1A and Fig. 2). This reaction was chosen as it is effectively entropy driven, proceeds well in an aqueous environment and also has the additional features of ease of installation and absence of cross reactivity with enzymic functional groups.
We have already used the active site of electric eel acetylcholinesterase (AChE) to template the formation of a ~100 femtoMolar inhibitor (Fig. 1C). We prepared arrays of sub-site inhibitors based on tacrine and phenanthridinium motifs decorated with alkyl-azides and alkyl-acetylenes (Fig. 1B). The diversity of the system originates from chain length and triazole geometry. Binary combinations of building blocks incubated with AChE were screened using Desorption Ionization on porous Silicon (DIOS) mass spectrometry. Only one inhibitor was produced by AChE from the 98 initial possibilities, thus demonstrating the feasibility of this process.

Figure 1: A) The non-templated 1,3 dipolar cycloaddition reaction of an azide and an acetylene affords a ca. 1:1 mixture of triazole regioisomers. B) The array of building blocks used and AChE as a partially transparent blue surface with a ribbon representation of the protein and the azide (green carbons, blue nitrogens) and acetylene (green carbons, magenta alkyne group) building blocks modelled in the binding site. C) The only product produced by AChE is the best known non covalent inhibitor of the template protein.

 

Figure 2: Click the figure above to see an animated GIF of the AChE assisted synthesis of an AChE inhibitor. Docking simulations of the building blocks and the full length inhibitor were performed separately using the program AutoDock.

This ongoing project is being carried out in collaboration with Dr. K. Barry Sharpless’ Lab (TSRI), Dr. M.G. Finn’s Lab (TSRI) and Dr. P. Taylor's Lab (UCSD).

 

Prediction of the protein-protein interactions in the LFA-1 I-domain / ICAM-1 D1 complex. A drug design endeavor.

Integrins mediate cell-cell and cell-matrix adhesion processes. Both enhanced and deficient leukocyte adhesion result in pathological states, therefore the activation of leukocyte is strictly regulated. It is evident that interfering with adhesion events may lead to numerous potential clinical applications, including the treatment of cancer, graft rejection, as well as chronic and acute infections. Although a subject of intense research, a detailed understanding of many of the molecular mechanisms involved in integrin mediated cell adhesion remains unclear.
Our program has two main objectives. Firstly, we aim to better understand the mechanism of integrin activation and secondly, we will apply this knowledge to develop possible therapeutics. We will focus our studies on the interactions of the Leukocyte Function associated Antigen-1 (LFA-1, aLb2) and its cellular ligands, namely the Intercellular Adhesion Molecules (ICAMs), which are the key to the tight adhesion of most activated leukocytes to the endothelial wall prior to trans-endothial migration. All efforts to date to inhibit the activity of the LFA-1 I-domain have focused on the closed "inactive" conformation. For the first time we will target both the closed as well as the biologically relevant open conformation of the LFA-1 I-domain for the development of therapeutic agents.
The integrin family of proteins consists of more than 20 ab heterodimeric membrane-bound glycoproteins, with 16 a- and 8 b-subunits. The aL in LFA-1 is one of nine integrin a-subunits (a1, a2, a10, a11, aD, aE, aL in LFA-1, aM in Mac-1, aX) that contain an inserted-domain (I-domain) of ~190 amino acids. The integrin I-domain is homologous to the von Willebrand factor A-domain and adopts the Rossmann fold. The I-domain is located towards the N-terminus of the a-subunit and plays a central role in ligand binding.
Crystallographic studies indicate that conformational changes (closed ß open) within the a2 and Mac-1 I-domains result in activation and subsequent ligand adhesion. To date no high-resolution structure of the open LFA-1 I-domain is available, although the existence of the two states is supported by NMR titration data of this protein with the ICAM-1 D1/D2 domains. The implication that the conformational changes accompanying the activation of the LFA-1 I-domain are parallel to those observed in the a2 and Mac-1 I-domains is the centerpiece of our working hypothesis.
We have modeled the structure of the LFA-1 I-domain in the open conformation using the Mac-1 I-domain as a template (Fig. 3). The 3-dimensional model was successfully used in docking studies with the ICAM-1 D1 domain. The key interaction of Glu 34 in the ICAM-1 D1 domain with the I-domain Metal Ion Dependent Adhesion Site (MIDAS) was reproduced by our calculations and the nature of the protein-protein interface is suggested (Fig. 4). The prediction of the open LFA-1 I-domain represents our structural starting point for the rational design of LFA-1 agonists and antagonists, as well as a template to experimentally express recombinant LFA-1 I-domain in the open conformation.

Figure 3: Ribbon diagrams of A) The superposition of Mac-1 I-domain in the closed (1JLM) and in the open conformation (1IDO) shown in orange and magenta respectively. B) The superposition of LFA-1 I-domain in the closed (1LFA) and in the modeled open conformation shown in cyan and green respectively. Mn+2 bound to the close conformation of the proteins is represented by a pink sphere, whereas Mg+2 bound to the open conformation is represented by a magenta sphere. Helices are labeled to illustrate changes between the closed and open conformations.

Figure 4: Docking of ICAM-1 D1-domain and LFA1 I-domain.

Many integrins recognize short peptide sequences, e.g. the RGD tripeptide, as well as other small molecules, e.g., lovastatin. These compounds could in principle inhibit integrin function and represent prototypes of a novel class of therapeutics. Lead compounds will be selected by searching chemical and oligopeptide databases including specially tailored libraries as well as from rational design of small molecule inhibitors. The integrin inhibitors will target both the I-domain MIDAS in the open conformation and the regulatory allosteric site at the C-terminal a7 helix in the closed conformation. Docking algorithms will be used to detect the potential best binders. Selected molecular candidates will then be synthesized and tested for their activity in cell adhesion assays. The results arising from structural and biological studies will provide the basis for novel models of integrin binding in an iterative design cycle.

This ongoing project is being carried out in collaboration with Dr. G. Legge (U. of Houston, Texas), Y. Takada (TSRI) and A. Olson (TSRI).

Multiple Reactive Immunization.

Catalytic antibodies presently represent the most successful class of tailor made macromolecules with enzyme-like properties. Immunization with stable transition-state analogs provides a good starting point for eliciting enzyme-like systems, but it is clear that this approach alone will not lead to very efficient catalysts. Using the reactive immunization approach, immunization is carried out with reactive compounds in order to promote a chemical reaction during the binding of the antigen to the antibody. Later, the same reaction becomes part of the mechanism of the catalytic event. It is reasonable to believe that the next major step in antibody catalysis will be the rational design of antibodies bearing more than one catalytic group in their binding sites. Ideally, the designed hapten should consist of two detachable parts: one carrying a reactive center and another one bearing a second tacit reactive center and also the carrier protein. Antibodies that bind to the hapten hold the promise of being efficient catalysts for the first reaction and in addition will present a second nucleophilic group within the active site. Following this concept, three haptens containing a phosphorous reactive center and a latent quinone methide were designed. The antibodies generated via double (or multiple) reactive immunization are expected to present more than one nucleophilic catalytic group in their binding sites and prove themselves be very effective catalysts for the hydrolysis of a wide range of organophosphate (phosphodi- and triesters) and organophosphonate compounds. Antibodies catalyzing the hydrolysis of toxic organophosphorous compounds will be of pharmacological importance. RNA and DNA catalytic cleavage will also be attempted. The antibodies will also be tested as catalysts for ester and amide (peptide bonds) hydrolysis.

Figure 5

Design and Total Synthesis of Artificial Enzymes.

During the last decade the rational design of molecules with enzyme-like properties emerged as a powerful tool to experimentally test general theories about molecular interactions and postulated catalytic mechanisms. This work focuses on the design and total synthesis of artificial enzymes via a "chemical ligation" approach using bovine cellular retinoic acid binding protein (CRABP) and 4-oxalocrotonate tautomerase (4-OT) as molecular scaffolds. The folded conformation of our synthetic polypeptides is expected to resemble that of the corresponding wild type protein. Natural and non-natural amino acids having desired catalytic and binding properties will be strategically introduced in the primary structure of the synthetic proteins to function as catalytic groups organized in a pre-determined three-dimensional shell. In the present study, the decarboxylase and aldolase activities of the amine groups of synthetically mutated residues within the protein’s binding site will be determined. The pKa of the key amino groups is expected to be perturbed by the hydrophobic environment of the binding site of the protein, accounting for the catalytic activity. This rather simple strategy is supported by previous reports on the catalytic activity of aldolase antibodies and aldolase class I enzymes. Calculated 3-D models endorse the feasibility of the proposed aldol condensation reactions catalyzed by the synthetic CRABP as well as its expected folding. Furthermore, the physico-chemical properties of functional groups within the hydrophobic pocket of a synthetic protein will be determined by the modification of singular structural elements. This project will evaluate the effect of substituents and variable chain length on the pKa of a side chain key amino group within the binding site of the artificial enzymes. In a progressive re-design cycle of the protein’s binding site, the kinetic and stereochemical control of the catalyzed transformations will be improved. In good time, the synthetic approach is expected to shed light on catalytic mechanisms and prove useful in the creation of novel catalysts.
The wild type 4-OT presents an N-terminus Pro residue in each of its six binding sites. We successfully introduced novel decarboxylase activity in this isomerase enzyme by rational design and total synthesis of the mutant protein catalyst. Interestingly, both the wild type and the mutants are efficient aldolases. The nature of the performed mutations also has implications in enzyme evolution.

Figure 6

This ongoing project is being carried out in collaboration with Dr. E. Keinan’s laboratory (Technion and TSRI), and Dr. P.E. Dawson’s laboratory (TSRI).

Grynszpan Laboratory: Home | Research | Personnel | Publications | Links | Gallery