Giardia gene predicts a 183 kDa nucleotide-binding head-stalk protein

Jonathan Marshall 0 David V. Holberton 0 0 Department of Life Science, Nottingham University , Nottingham, NG7 2RD, UK *Author for correspondence SUMMARY Previously described extended proteins from the cytoskeleton of Giardia lamblia (b -giardin, median body protein) have been found to be segmented coiled coils with regular structural repeat patterns in their amino acid sequences. Screening a l ZAPII library derived from Giardia genomic DNA with an antibody directed against a 34 103 Mr giardin isoform selected a gene encoding a much larger polypeptide chain (HPSR2), the sequence of which was determined by chromosome walking the open reading frame. The complete gene has been cloned and expressed as a recombinant protein of 183 103 Mr. The predicted amino acid sequence of the protein has identifiable features suggesting that it might be a motor protein with an amino-terminal hydrolytic domain attached to a long coiled coil stalk. The Coiled coil structures are of considerable importance in the cytoskeleton. The tendency for parallel amphipathic a -helices to lock together by hydrophobic side chain docking (Crick, 1953) provides a means whereby rigid elongated molecules suitable for struts and filaments can be generated. Major coiled coil families (tropomyosins, myosins, intermediate filament (IF) proteins) were recognised early in fibrous tissues. Recent sequencing projects have discovered further notable examples in the cytoplasm (e.g. kinesin (Yang et al., 1989), desmoplakin (Green et al., 1990), tektin, (Norrander et al., 1992)), and concerned with division events in the nucleus of yeast (Alani et al., 1989; Mirzayan et al., 1992; Kolling et al., 1993) and mammalian cells (Yang et al., 1992). In the case of the best characterised motor proteins, kinesin and sarcomeric myosin heavy chains, long coiled coil segments are seen to anchor or transmit the forces produced at the hydrolytic domain to enable useful mechanical work to be done. We are interested in the coiled coil proteins expressed in the zooflagellate Giardia, which may be the most primitive of known eukaryotic organisms (Sogin et al., 1989). Previously, we have sequenced two coiled coil proteins from different loci in the microtubule cytoskeleton (Holberton et al., 1988; Marshall and Holberton, 1993). Their genes were isolated by expression screening cDNA and genomic libraries with monospecific polyclonal antibodies to cytoskeleton proteins. Coiled coil amino acid sequences are recognised by characteristic heptapeptides (heptads) with alternating 3,4 residue spacing of presumed head domain is 211 residues and contains a Ploop sequence conserved in purine nucleotide-binding proteins. The remaining 1409 amino acids mainly make up a region of heptad repeats such as in myosin or the kinesin stalk, ending in a small (67 amino acids) carboxy-terminal domain. Fourier analysis of the predicted stalk shows the presence of a strong physical repeat created by regular heptad phase changes dividing the coil into segments of 25 residues. This structure most closely resembles the smaller microtubule-associated median body protein which has segments of 24 residues. apolar side chains. The two Giardia proteins are unusual in that their heptad series are regularly interrupted by extra residues which modify the apolar Fourier period. This feature has the effect of dividing the coils into successions of similar segments. Beta-giardin is one of a large number of ~ 30 kDa proteins (giardins) found in the cytoskeleton (Clark and Holberton, 1988; Peattie et al., 1989). It is a small coiled coil with segments of four heptads followed by a skip residue (Holberton et al., 1988). The second protein is a larger (101 kDa) three-domain protein from the median body bundle of microtubules. Its long central rod domain has segments of 24 residues created by periodically reversing the phase of the apolar repeat (Marshall and Holberton, 1993). Such structural patterns might have some role in the association of these rod proteins with microtubules. A structural homologue of b giardin, somewhat dissimilar in sequence but with an identical heptad pattern, has now been found in fibres of the flagellar basal apparatus of the green alga Spermatozopsis similis (Weber et al., 1993). We have been using rat antibodies to ~ 30 kDa polypeptides from the giardin isoform cluster to screen for other b -giardinlike coiled coils. One antiserum to a component more basic than b -giardin bound strongly to the immunising isoform and, at a lower level of reactivity, to b -giardin itself. Used to screen a genomic l ZAPII expression library (Marshall and Holberton, 1993), the antibodies isolated two clones with the same insert. The expressed gene is an open reading frame (ORF) much longer than a ~ 30 kDa polypeptide, but which translates to an amino acid heptad sequence. The complete ORF has now been resolved as encoding a 183 kDa head-stalk type of protein. We report this sequence in the present paper. By homology, the fold of the head domain incorporates a mononucleotide phosphate binding site, as in a myosin or kinesin heavy chain. However, similarity to these two motor proteins does not extend beyond the active site in the head domain and the stalk heptads. The head domain is also smaller, about two thirds the size of the kinesin head. Like Giardia median body protein, the stalk sequence is regularly patterned by heptad phase changes, but in this case the coil segments so formed are of 25 rather than 24 residues. MATERIALS AND METHODS Primary screening antibodies were polyclonal rat antisera raised to specific Giardia cytoskeleton proteins, as previously described (Holberton et al., 1988; Marshall and Holberton, 1993). Proteins were isolated by two-dimensional sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) of detergent-insoluble cytoskeletons dissolved in 9 M urea (Clark and Holberton, 1988). One antiserum (R1 antibodies) screened the clones described in this paper from a genomic DNA expression library in l ZAPII constructed previously (Marshall and Holberton, 1993). The R1 immunogen was a prominent 34 kDa spot running in the most basic region (pH ~6.0) of the focusing gradient, well separated from the main giardin isoform cluster (visible in Fig. 1a of Marshall and Holberton, 1993). Library fragments were mung bean nuclease fragments of genomic DNA from Giardia lamblia (Portland-1 strain). Inserts tend to coincide with coding regions because of the preference of the enzyme for A-T rich intergene sequences (McCutchan et al., 1984). For immunoscreening, amplified library samples were plated on E. coli XL1-Blue, and lacZ expression induced with isopropyl-b -D-thiogalactoside (IPTG) (Huynh et al., 1985). Nitrocellulose filter lifts were blocked (5% Marvel dried milk powder in Tris-buffered saline: 10 mM Tris-HCl, 154 mM NaCl, pH 7.4, with 0.05%(v/v) Tween-20 (TBS/T)) before 12-16 hours incubation in primary antibodies in TBS/T with 5-10% (...truncated)