Introduction

Calcium is an essential component in the biomineralization of teeth, bones, and shells, as well as a second messenger regulating cellular processes such as cell division and growth, secretion, ion transport, and muscle contraction (Da Silva & Williams, 1991 ; Linse & Forsen, 1995). Calcium ions regulate many different biological processes by binding to proteins with different affinity. The binding of calcium to proteins leads to an increase of stability and changes in conformations of the calcium binding proteins. There are many calcium-binding proteins located in different cellular compartments, such as extracellular, intracellular, and the nucleus (Fig. 1). The mediation and regulation of Ca(II)-dependent functions are fulfilled by the change of calcium concentrations in the different cellular environments. The understanding of the molecular basis of diseases caused by the overloading of calcium, the disruption of calcium binding sites of proteins, and the mechanism of calcium-modulated signal transduction requires the establishment of the principles for the affinity of calcium binding proteins (Kawaski & Kretsinger, 1995; Schafter & Heizman, 1996).

The rational design of novel proteins has been shown to be a powerful approach for the identification of the key factors involved in protein functions, to study structure and protein folding, and to establish principles for the prediction of the function of the proteins (Bryson et al., 1995; Hellinga, 1998). In addition, protein design is a way to construct new biomaterials, sensors, catalysts and pharmaceutical drugs. Rapid progress has been made in designing metal ions such as zinc, cobalt and nickel in proteins (Regan, 1995; Shi et al., 1996; Lu & Valentine, 1997). However, little progress has been made in the design of Ca(II) binding proteins. So far, only a handful of designed Ca(II) binding studies have been published. All of them are based on the transfer of a Ca(II) binding loop into a structural analog, such as the surface-exposed Ca(II) binding loop from thermolysin into neutral protease (Toma et al., 1991), and that from a-lactalbumin into egg white lysozyme (Aramini et al., 1992; Regan, 1993). The de novo design of calcium binding sites, however, has not been achieved. Designing calcium-binding sites in non-calcium-binding proteins will deepen our understanding of the calcium-protein interaction without interfering with the metal-metal interaction. This kelps us to understand the key determinants for calcium-binding affinity, and brings us closer to the design of novel calcium-modulated proteins with specifically desired functions.

Figure 1. Three-dimensional structures of the intracellular EF-hand calcium-binding proteins calmodulin (3cln), parvalbumin (5cpv), and calbindinD9t (4icb).

One of the major barriers for the design of calcium binding sites in proteins comes from the complexity and irregularity of calcium binding sites. Detailed surveys (Strynadka & James, 1989; da Silva & Williams, 1991; Glusker, 1991; Falke et al., 1994; Kawasaki & Kretsinger, 1995; Nelson & Chazin, 1998) of more than one hundred known Ca(II)-binding proteins and several hundreds of Ca(II)-small molecule complexes have revealed that almost all of the ligands at Ca(II) binding sites in proteins are oxygen atoms (Fig. 2). The side chains of Asp, Glu, and Asn, the carbonyl groups from the main chain, and water from solvents (Glusker, 1991; Falke et al., 1994) all provide oxygen atoms to chelate calcium. The deviations in the Ca-0 bond length are from 2.2 to 2.8 A with a mean Ca(II)-ligand distance of 2.42 A (Glusker, 1991; Falke et al., 1994). The coordination number varies from 3 to 9 while 6 or 7 coordination atoms are commonly found in proteins. The most common calcium-binding site has a pentagonal bipyramid or distorted octahedral geometry (Fig. 2). Ca(II)-binding sites with coordination numbers of seven or eight tend to employ more bidentate and water ligands, possibly due to ligand crowding around the ions.

A class of evolutionarily related intracellular calcium binding proteins (EF-hand proteins) have been shown to have strong homology both in the protein sequence, protein frame, and calcium binding sites (Kuboniwa et al., 1995; Finn et al., 1995; Zhang et al., 1995; Nelson & Chazin, 1998) (Fig. 1). To date, there are more than 500 appearances of EF-hand proteins in the protein and gene data banks. EF-hand proteins contain a variety of subfamilies, such as parvalbumin (Parv), troponin C (TnC), calmodulin (CaM), sarcoplasmic calcium binding protein, the essential and regulatory light chains of myosin, calbindinD9K (CBD), the S100, and VIS subfamilies (Linse & Forsen, 1995; Kawasaki & Kretsinger, 1995). All the EF-hand motifs we know of consist of a highly conserved loop flanked by two helices (helix-loop-helix) (Kretsinger & Nockolds, 1973), which can be further divided into classic EF-hand motifs and pseudo-EF-hand motifs. The residue numbers in the calcium-binding loop are 12 and 14 for the classic EF-hand and pseudo-EF-hand motifs, respectively. For a classic EF-hand motif (Fig. 2), seven oxygen atoms from the sidechains of Asp, Asn, and Glu, the main chain, and water at the loop sequence positions of 1, 3, 5, 7, 9, and 12 coordinate the calcium ion in a pentagonal bipyramidal arrangement. In a typical geometry, position 1 of the calcium-binding loop is always Asp and the side-chain of Asp serves as a ligand on the x-axis. The -x-axis (position 9) is filled with a bridged water molecule connecting the sidechain of Asp, Ser and Asn (Fig. 2) (Kretsinger and Nockolds, 1973; Strynadka and James, 1989). Axis -z is shared by the two carboxyl oxygen atoms of a glutamate side chain at position 12 that binds in a bidentate mode to Ca(II). Glu is used predominantly (92%) as a bidentate ligand for both classic and pseudo EF-hand motifs in all intracellular calcium-binding proteins (Falke et al., 1994; Kawasaki & Kretsinger, 1995). For the calcium-binding loop of a pseudo-EF-hand motif, four oxygen atoms from the main-chain carbonyl groups at sequence positions 1,4, 6, and 9 provide ligands for the calcium ion with a coordination geometry very similar to that of a classic EF-hand motif (Svensson et al., 1992; Linse & Forsen, 1995). Two EF-hand motifs are almost always associated in the same domain of a protein to yield highly cooperative calcium binding systems (Falke et al., 1994; Kawasaki & Kretsinger, 1995). Calmodulin, for example, contains four EF-hand calcium-binding sites in two domains (Fig. 1). CalbindinD9k contains a pseudo EF-hand motif for calcium binding site 1, while it contains a classic EF-hand motif for site 2. Although EF-hand proteins have shown strong homology in primary sequences and similarity in the protein frame and calcium binding sites, their cellular functions, especially the response to calcium binding, are extremely diverse. Calcium binding tightly regulates the functions of trigger-like EF-hand proteins, such as calmodulin and troponin C, since calcium binding to trigger proteins leads to large changes in their conformations and the exposure of their hydrophobic surfaces. On the other hand, calcium binding does not result in a significant conformational change of nontrigger or buffer proteins, such as CalbindinD9K, whose functions are proposed to maintain proper cellular calcium concentrations. The origins for the different responses to calcium binding between trigger and non-trigger proteins are not clear (Nelson & Chazin, 1998). It is necessary to investigate factors contributing to both classes of EF-hand proteins.

Figure 2. The calcium-binding site of calmodulin with a pentagonal bipyramidal geometry (left). Strategy of Dezymer program (right). The first residue located in the calculation (called anchor) defines the relative position of the calcium atom to the protein backbone and is used as a starting point to construct a calcium-binding site. After attaching the anchor residue to the backbone of the protein along the protein sequence, the calcium-binding geometry or positions of other ligands are then defined around the anchor.

Figure 2. The calcium-binding site of calmodulin with a pentagonal bipyramidal geometry (left). Strategy of Dezymer program (right). The first residue located in the calculation (called anchor) defines the relative position of the calcium atom to the protein backbone and is used as a starting point to construct a calcium-binding site. After attaching the anchor residue to the backbone of the protein along the protein sequence, the calcium-binding geometry or positions of other ligands are then defined around the anchor.

The rational design of calcium binding sites into proteins requires several capabilities for a computer algorithm such as placing large coordination numbers (e.g. 7 and 8), incorporating different ligand types and accommodating the irregularity of metal binding sites. Several computer programs have been developed for the design of metal binding sites in proteins. Algorithms based on the analysis of hydrophobicity and charge valences of metal binding sites in protein data banks have been developed in Eisenberg's group and in Cera's lab (Yamamshita et al., 1990; Nayal & Cera, 1994). METAL, SEARCH has been established for the de novo design of tetrahedral Zn(II) sites in proteins (Clarke & Yuan, 1995). This program has been successfully tested in the design of sites in a4 and pi using His and Cys as ligands (Regan & Clarke, 1990; Klemba et al., 1995).

In this work, we report our studies to establish structural parameters for identifying and designing calcium-binding proteins by the computer algorithm Dezymer. Different metal sites with low ligand numbers such as zinc fingers, blue copper, and [Fe4S4] clusters have been successfully constructed in the hydrophobic core of Escherichia Coli thioredoxin by Hellinga and his colleagues (Hellinga & Richards, 1991; Hellinga, 1998). Here, we have used three EF-hand proteins, calmodulin, parvalbumin and calbindinD9K, as model systems to test if this computer program has the ability to accommodate the intrinsic variability of ligand geometry and length of Ca(II)-ligand oxygen atom. We demonstrate that oxygen atoms from mainchain carbonyl and water can be implanted as ligands. Both classic and pseudo EF-hand binding sites in natural EF-hand proteins can be relocated using the parameters from the geometry of an ideal pentagonal bipyramid. We further demonstrate that the natural sites have the smallest deviation from the ideal pentagonal bipyramidal geometry.

0 0

Post a comment