B Calcium Binding EF Hands

These calcium binding motifs were first identified by Kretsinger using the crystal structure of the Carp calcium binding protein parvalbumin, who designated them EF hands after the proximal E- and F-helices of parvalbumin (Tufty & Kretsinger 1975). The liganding loop consists of twelve amino acids, of which five have carboxyl or hydroxyl groups in their side chains, that are precisely spaced to coordinate the calcium ion, the loop is flanked by two alpha helix loops (for review see Heizmann & Hunziker 1991 and Kretsinger 1980).

An alignment of various calcium binding sequences displaying standard EF hand loops indicated that the first aspartate (D) amino acid at position one is particularly critical for the formation of the liganding loop. Mutation of this residue abolishes calcium binding and function of cardiac troponin C (Putkey et al 1989) and yeast calmodulin (Geiser et al

1991). Comparison of the transgelin sequence with the EF hand structure indicates the existence of a region that possesses a number of these critical residues. However, the first and most critical aspartate exists as a lysine (K) that would be unable to provide the correctly aligned oxygen atom necessary for calcium ion liganding. This would suggest that this protein would be unable to bind calcium. Gel-overlay methods using radioactively labelled calcium (^^Ca) (Maruyami et al 1984) confirms that purified transgelin is unable to bind calcium (D Lawson unpublished). ‘In vitro’ experiments using purified transgelin and actin in a falling-ball viscometry system and rebinding to detergent extracted cells similarly suggest that the gelling and binding function of transgelin are independent of calcium concentrations (Shapland et al 1993).

The EF-hand structure represents an ‘ancient’ genetic motif. Analysis of sequence conservation in different species indicates that the structure had been multiply duplicated by a time considerably preceding the divergence of the yeast and vertebrate evolutionary branches over one thousand million years ago (Wilson et al 1988). It is not currently known at what point in evolution, if any, transgelin may have possessed and then lost its ability to bind calcium. Sheep, rat and human sequences have been determined for this region and suggest that these transgelins would similarly be unable to bind calcium.

Alternatively, it is possible that the motif found in transgelin is a conserved sequence that preceded the evolution of the EF hand structure. This will be tested by the complementary approaches of cloning and sequencing of transgelin homologues from organisms such as yeast and the use of the ^^Ca gel overlay technique on protein extracts from representative organisms from the various evolutionary branches.

A second class of proteins exemplified by the lipocortins (p36) contain a conserved segment of approximately seventy amino acid residues that might include the site for association with phospholipids. Although there is no significant sequence similarity within this region with the EF-hand motif, these proteins can bind calcium and this allows them to interact with phosholipids and cellular membranes in a calcium dependent manner using an unknown mechanism (reviewed in Heizmann & Hunziker 1991; Crompton et al 1988). While transgelin displays no detectable homology with these proteins it remains possible that modulation of protein behaviour by calcium can occur without the involvement of an EF-hand and may only be detected ‘in vivo’ where multiple protein interactions act synergistically (eg the gelsolin SI fragment traps calcium between itself and actin; Way et al 1992b).

4.8 Hydropathy Plots

The measure of hydropathy is intended to reflect the ‘contrary tendencies of hydrophilicity and hydrophobicity’, namely the polarity of different regions. When proteins fold in aqueous solution the most stable state ought to be that in which the maximum number of polar groups are on the surface and in contact with water, while at the same time the maximum number of non-polar side chains are buried away from the surface (Doolittle 1986). Hydropathic values at each amino acid are calculated by statistically averaging the index values for the six adjacent amino acids. This iterative process is performed moving along one amino acid at a time until overlapping values have been obtained for the entire protein (the resulting range is usually +5 to -5) and can be displayed graphically (Kyte & Doolittle 1982). The hydropathicity plot for transgelin was examined for regions of interest and areas of the sequence involved in electrostatic or alpha helical interactions with other proteins should, in theory, be present in hydrophilic stretches, although this has been shown to not be the case for the 27 amino acid actin binding region in ABP120 (Bresnick et al 1991).

The transgelin plot displays no highly hydrophobic charge free regions characteristic of transmembrane or membrane associated regions. Proteins that are to be ‘exported’ from the cell commonly contain a highly hydrophobic N-terminal leader sequence, that is proteolytically removed during or after ‘export’ (Alberts et al 1994), conversely transgelin possesses a relatively hydrophilic N terminus that would preclude its secretion or membrane association. This correlates with the non-membranous distribution of C4 monoclonal antibody staining seen in mesenchymal cells by immunofluorescence (Shapland et al 1988).

The highly acidic (negatively charged) region (amino acids 23-29, DEELEE) exists in a highly hydrophilic area and may therefore be ‘available’ for electrostatic interactions with other proteins. The highly basic (positively charged) region (amino acids 154-161, KKAQEHKR discussed Section 4.6.1C(i)) is also present in a highly hydrophilic stretch and so similarly would be ‘available’ for interactions with other proteins such as actin. This positive region is thought to be the sequence affected in falling ball viscometry assays in which the inclusion of polyphosphate ions (Muhlrad 1991) specifically abolishes the ability of transgelin to gel actin (Shapland et al 1993).

The putative phosphorylation sites (amino acids 4-7, KGPS; 181-186, SNRGAS) are in moderately hydrophilic regions that would be accessible to kinase enzymes. Phosphorylation of these sites may alter the hydropathy of the region and induce a conformational change in the surrounding area, potentially in the cluster of positive charges (154-161, KKAQEHKR) modifying the ability of transgelin to interact with actin as seen with other proteins whose activity is modulated by phosphorylation (see Section 4.7.1 A above).

4.9 Secondary Structure Predictions

Various algorithms predicting the secondary structure of tri- to penta-peptides based upon the distribution of amino acids in proteins with known three dimensional structures are available (for example Chou & Fasman 1974 and Gamier et al 1978). As these methods involve a combination of empirical and theoretical considerations it is often necessary to run two or more predictive programs to ensure the results are comparable (Nishikawa 1983; reviewed in Sternberg 1992; Benner & Gerloff 1993). While these results are subject to significant errors and should be treated with caution (Kabsch &

Sander 1983; Kabsch & Sander 1984) they can provide valuable insights into the possible structure of globular proteins. The Joint prediction suite used to analyse transgelin utilises eight independent methods to minimise systematic errors and thus represent a reasonably accurate structural map. These systems predict that the majority of the transgelin protein will exist as an alpha-helix and so adopt an extended rod-like conformation as seen for tropomyosin (Ruiz-Opazo & Nadal-Ginard 1987). While beta sheets provide considerable stmctural stability most interactions between proteins often occur by the co-alignment of alpha helices (Kabsch et al 1990; Milligan et al 1990). It has been suggested that the actin binding sites in hisactophilin, cofilin, villin and profilin are all located in regions with high alpha-helical potential (for review see Vandekerckhove and Vancompemolle 1992) and it may therefore be of significance that three potentially important regions (the two charged clusters and the LKAAEDY motif; amino acids 22-29,97-104 and 154-151) (discussed in Sections 4.6.1C and 4.12) occur in regions likely to exist as alpha helices.

4.10.1 Sequence Conservation

Comparison of the derived sheep aorta cDNA and rat small intestine cDNA sequences for transgelin demonstrated 99% amino acid conservation. Similar comparison of the amino acid sequences obtained from the purified sheep transgelin protein with the sheep cDNA derived amino acid sequence indicated 96% conservation. While this marginally lower value may reflect errors in the cDNA sequence introduced during PCR amplification, this is unlikely since these reactions were performed using Pfu polymerase (Stratagene) known to display a proof-reading/ editing and so were more likely to reflect the technical limitations of direct amino acid sequencing, particularly with short peptides isolated in limiting quantities which serve to distort the overall accuracy. Nevertheless, the very high homology between these sequences confirms that the clones encode the rat and sheep homologues of transgehn respectively. In view of this the protein encoded by the rat cDNA clones would be expected to behave hke transgelin purified from sheep aorta both in vitro and in vivo .