Colicin E1 binding to membranes: time-resolved studies of spin-labeled mutants.

To investigate the mechanism of interaction of the toxin colicin E1 with membranes, three cysteine substitution mutants and the wild type of the channel-forming fragment were spin labeled at the unique thiol. Time-resolved interaction of these labeled proteins with phospholipid vesicles was investigated with stopped-flow electron paramagnetic resonance spectroscopy. The fragment interacts with neutral bilayers at low pH, indicating that the interaction is hydrophobic rather than electrostatic. The interaction occurs in at least two distinct steps: (i) rapid adsorption to the surface; and (ii) slow, rate-limiting insertion of the hydrophobic central helices into the membrane interior.

A single tryptic fragment of colicin E1 can form an ion channel: stoichiometry confirms kinetics.

The molecularity of the ion channel formed by peptide fragments of colicin has taken on particular significance since the length of the active peptide has been shown to be less than 90 amino acids and the lumen size at least 8 A. Cell survival experiments show that killing by colicin obeys single-hit statistics, and ion leakage rates from phospholipid vesicles are first order in colicin concentration. However, interpretation in molecular terms is generally complicated by the requirement of large numbers of colicin molecules per cell or vesicle. We have measured the discharge of potential across membranes of small phospholipid vesicles by following the changes in binding of potential sensitive spin labeled phosphonium ions as a function of the number of colicin fragments added. Because of the sensitivity of the method, it was possible to reliably investigate the effect of colicin in a range where there was no more than 0.2 colicins per vesicle. The quantitative results of these experiments yield a direct molecular stoichiometry and demonstrate that one C-terminal fragment of the colicin molecule per one vesicle is sufficient to induce a rapid ion flux in these vesicles. In addition, the experiments confirm earlier findings that the colicin fragments do not migrate from one vesicle to another at pH 4.5. Similar results are obtained with large unilamellar vesicles.

Site-directed mutagenesis of colicin E1 provides specific attachment sites for spin labels whose spectra are sensitive to local conformation.

Colicin E1 is an E. coli plasmid-encoded water-soluble protein that spontaneously inserts into lipid membranes to form a voltage-gated ion channel. We have employed a novel approach in which site-directed mutagenesis is used to provide highly specific attachment points for nitroxide spin labels. A series of colicin mutants, differing only by the position of a single cysteine residue, were prepared and selectively labeled at that cysteine. A hydrophilic sequence (398-406) within the C-terminal domain of the water-soluble form of the protein was investigated and exhibited an electron paramagnetic resonance (EPR) spectral periodicity strongly suggesting an amphiphilic alpha-helix. After removal of the N-terminus of the protein with trypsin, the spectra for this sequence indicate increased label mobility and a more flexible structure.

A Purkinje cell differentiation marker shows a partial DNA sequence homology to the cellular sis/PDGF2 gene.

To search for genes involved in determining the morphology of individual neuronal types, a cDNA library was constructed from postnatal day 13 mouse cerebellum. From this library, 2 clones, L7 and L19, were isolated by a differential hybridization procedure and shown by in situ hybridization to be Purkinje cell-specific within the cerebellum. Both RNAs appear between postnatal days 4 and 8 and continue into adulthood, coinciding with terminal differentiation of the Purkinje cells. L7 seems to be expressed exclusively in the cerebellum, whereas L19 is expressed throughout the brain. Consistent with the RNA localization, L7 protein is found only in the cerebellum and is confined to the Purkinje cells. The L7 amino acid sequence has been deduced from the cDNA sequence, and a pseudo-repeat within the L7 protein sequence is homologous to the amino acids sequence in the primary translation product of the gene for human sis/PDGF.

A very short peptide makes a voltage-dependent ion channel: the critical length of the channel domain of colicin E1.

Cleavage of colicin E1 molecules with a variety of proteases or with cyanogen bromide (CNBr) generates COOH-terminal fragments which have channel-forming activity similar to that of intact colicin in planar lipid bilayer membranes. The smallest channel-forming fragment obtained by CNBr cleavage of the wild-type molecule consists of the C-terminal 152 amino acids. By the use of oligonucleotide-directed mutagenesis, we have made nine mutants along this 152 amino acid peptide, in which an amino acid was replaced by methionine in order to create a new CNBr cleavage site. The smallest of the CNBr-cleaved C-terminal fragments with channel-forming activity, in planar bilayer membranes, was generated by cleavage at new Met position 428 and has 94 amino acids, whereas a 75 amino acid peptide produced by cleavage of a new Met at position 447 did not have channel activity. The NH2-terminus of the channel-forming domain of colicin E1 appears therefore to lie between residues 428 and 447. Since, however, the last six C-terminal residues of the colicin can be removed without changing activity, the number of amino acids necessary to form the channel is 88 or less. In addition, the unique Cys residue in colicin E1 was replaced by Gly, and nine mutants were then made with Cys placed at sequential locations along the peptide for eventual use as sulfhydryl attachment sites to determine the local environment of the replaced amino acid. In the course of making 21 mutants, eight charged residues have been replaced by uncharged Met or Cys without changing the biological activity of the intact molecule. It has been proposed previously that the conformation of the colicin E1 channel is a barrel formed from five or six alpha-helices, each having 20 amino acids spanning the membrane and two to four residues making the turn at the boundary of the membrane. Our finding that 88 amino acids can make an active channel, combined with recently reported stoichiometric evidence that the channel is a monomer excludes this model and adds significant constraints which can be used in building a molecular model of the channel.