Identification and characterization of Kelch homologs

Microtubules regulate actin-based processes such as cell migration and cytokinesis, but the molecular mechanisms at work are not yet understood. In the fission yeast Schizosaccharomyces pombe, microtubule plus ends regulate cell polarity in part by transporting the kelch repeat protein tea1p to cell ends. This study identifies tea4p, a SH3 domain protein that binds directly to tea1p. Like tea1p, tea4p localizes to growing microtubule plus ends and to cortical sites at cell ends, and it is necessary for the establishment of bipolar growth. Tea4p binds directly to and recruits the formin for3p, which nucleates actin cable assembly. During 'new end take off' (NETO), formation of a protein complex that includes tea1p, tea4p, and for3p is necessary and sufficient for the establishment of cell polarity and localized actin assembly at new cell ends. These results suggest a molecular mechanism for how microtubule plus ends regulate the spatial distribution of actin assembly (Martin, 2005).

The possible homologs of tea1p and tea4p have generally not yet been well characterized. In S. cerevisiae, the nearest homologs of S. pombe tea1p, tea4p, and for3p are Kel1/Kel2p, Bud14p, and Bni1p, respectively. Although association between these factors has not been reported to date, mutant phenotypes and genetic interactions suggest that these proteins also function together to regulate cell polarity. In animal cells, the equivalents of tea1p and tea4p are less clear. A mammalian protein with some functional similarity (but low sequence homology) to tea4p is WISH/DIP. Like tea4p, this protein contains an N-terminal SH3 domain protein and interacts with mammalian formins, mDia1 and FHOD1, with an N-terminal region binding to FHOD1. Many kelch proteins regulate cytoskeletal processes in animal cells (Adams, 2000). Of note, the kelch repeat protein gigaxonin associates with MTs in neurons, Drosophila Kelch organizes ring canals, and Keap1 associates with the mid-body, adherens junctions, and focal adhesions and is required for cytokinesis. Further study of these proteins and their possible interactions will be a key part of understanding the molecular principles of cell morphogenesis (Martin, 2005).

Six independent mutations in the Caenorhabditis elegans spe-26 gene cause sterility in males and hermaphrodites by disrupting spermatogenesis. Spermatocytes in mutants with the most severe alleles fail to complete meiosis and do not form haploid spermatids. Instead, these spermatocytes arrest with missegregated chromosomes and mislocalized actin filaments, endoplasmic reticulum and ribosomes. In spite of this arrest some of the nuclei and the organelles that normally transport sperm-specific components to the spermatid mature as if they were in spermatids. The spe-26 gene is expressed throughout the testis in both spermatogonial cells and spermatocytes. It encodes a 570-amino-acid polypeptide, which contains five tandem repeat motifs, each of approximately 50 amino acids. These repeats are similar in sequence to repeats in the Drosophila Kelch protein, in the invertebrate sperm protein scruin that cross-links actin filaments, as well as in the mouse and pox virus proteins. The functional importance of these repeat motifs is shown by the fact that five of the spe-26 mutations are in the tandem repeats, and one of the most severe mutations is a substitution in a highly conserved glycine. These results suggest that spe-26 encodes a cytoskeletal protein, perhaps actin binding, that is necessary to segregate the cellular components that form haploid spermatids (Varkey, 1995).

The acrosomal process of Limulus sperm is an 80-microns long finger of membrane supported by a crystalline bundle of actin filaments. The filaments in this bundle are crosslinked by a 102-kD protein, scruin present in a 1:1 molar ratio with actin. Recent image reconstruction of scruin decorated actin filaments at 13-Å resolution shows that scruin is organized into two equally sized domains bound to separate actin subunits in the same filament. The gene for scruin has been cloned and sequenced from a Limulus testes cDNA library. The deduced amino acid sequence of scruin reflects the domain organization of scruin: it consists of a tandem pair of homologous domains joined by a linker region. The domain organization of scruin is confirmed by limited proteolysis of the purified acrosomal process. Three different proteases cleave the native protein in a 5-kD Protease-sensitive region in the middle of the molecule to generate an NH2-terminal 47-kD and a COOH-terminal 56-kD protease-resistant domains. Although the protein sequence of scruin has no homology to any known actin-binding protein, it has similarities to several proteins, including four open reading frames of unknown function in poxviruses, as well as Kelch, a Drosophila protein localized to actin-rich ring canals. All proteins that show homologies to scruin are characterized by the presence of an approximately 50-amino acid residue motif that is repeated between two and seven times. Crystallographic studies reveal this motif represents a four beta-stranded fold that is characteristic of the 'superbarrel' structural fold found in the sialidase family of proteins. These results suggest that the two domains of scruin seen in EM reconstructions are superbarrel folds, and they present the possibility that other members of this family may also bind actin (Way, 1995b).

Scruin (alpha-scruin) is an actin bundling protein found in the acrosomal process of Limulus polyhemus sperm. A second scruin isoform has been cloned and sequenced from Limulus, beta-scruin, that is 67% identical to alpha-scruin. Northern and Southern analyses confirm that beta-scruin and alpha-scruin are encoded by distinct genes. The sequence of beta-scruin, like alpha-scruin, is organized into N- and C-terminal superbarrel domains that are characterized by a six-fold repeat of a 50 residue motif. Western analysis using rabbit polyclonal antisera specific for alpha- and beta-scruin indicate that beta-scruin, like alpha-scruin, is found in Limulus sperm but not blood or muscle. Both immunofluorescence microscopy and immunogold-EM localize beta-scruin within the acrosomal vesicle at the anterior of sperm but not in the acrosomal process. The function of beta-scruin in this membrane-bounded compartment that is devoid of actin is unknown. However, the location of beta-scruin together with the fact that it contains two putative beta-superbarrel structural folds, which are known to be catalytic domains in a number of proteins, suggests it may have a possible enzymatic role (Way, 1995a).

The cytoskeleton plays an important role in neuronal morphogenesis. A novel actin-binding protein, Mayven, has been identified and characterized that is predominantly expressed in brain. Mayven contains a BTB (broad complex, tramtrack, bric-a-brac)/POZ (poxvirus, zinc finger) domain-like structure in the predicted N terminus and 'kelch repeats' in the predicted C-terminal domain. Mayven shares 63% identity (77% similarity) with the Drosophila ring canal Kelch protein. Somatic cell-hybrid analysis indicates that the human Mayven gene is located on chromosome 4q21.2, whereas the murine homolog gene is located on chromosome 8. The BTB/POZ domain of Mayven can self-dimerize in vitro, which might be important for its interaction with other BTB/POZ-containing proteins. Confocal microscopic studies of endogenous Mayven protein reveal a highly dynamic localization pattern of the protein. In U373-MG astrocytoma/glioblastoma cells, Mayven colocalizes with actin filaments in stress fibers and in patchy cortical actin-rich regions of the cell margins. In primary rat hippocampal neurons, Mayven is highly expressed in the cell body and in neurite processes. Binding assays and far Western blotting analysis have demonstrated association of Mayven with actin. This association is mediated through the 'kelch repeats' within the C terminus of Mayven. Depolarization of primary hippocampal neurons with KCl enhances the association of Mayven with actin. This increased association results in dynamic changes in Mayven distribution from uniform to punctate localization along neuronal processes. These results suggest that Mayven functions as an actin-binding protein that may be translocated along axonal processes and might be involved in the dynamic organization of the actin cytoskeleton in brain cells (Soltysik-Espanola, 1999).

Interactions of Kelch homologs with actin

Frozen, hydrated acrosomal bundles from Limulus sperm were imaged with a 400 kV electron cryomicroscope. Segments of this long bundle can be studied as a P1 crystal with a unit cell containing an acrosomal filament with 28 actin and 28 scruin molecules in 13 helical turns. A novel computational procedure was developed to extract single columns of superimposed acrosomal filaments from the distinctive crystallographic view. Helical reconstruction was used to generate a three-dimensional structure of this computationally isolated acrosomal filament. The scruin molecule is organized into two domains that contact two actin subunits in different strands of the same actin filament. A correlation of Holmes' actin filament model to the density in the acrosomal filament map shows that actin subdomains 1, 2, and 3 match the model density closely. However, actin subdomain 4 matches rather poorly, suggesting that interactions with scruin may have altered actin conformation. Scruin makes extensive interactions with helix-loop-beta motifs in subdomain 3 of one actin subunit and in subdomain 1 of a consecutive actin subunit along the genetic filament helix. These two actin subdomains are structurally homologous and are closely spaced along the actin filament. This model suggests that scruin, which is derived from a tandemly duplicated gene, has evolved to bind structurally homologous but non-identical positions across two consecutive actin subunits (Schmid, 1994).

In the acrosomal process of Limulus sperm, the beta-propeller protein scruin cross-links actin into a crystalline bundle. To confirm that scruin has the topology of a beta-propeller protein and to understand how scruin binds actin, the solvent accessibility of cysteine residues in scruin and the acrosomal process were compared by chemical modification with (1,5-IAEDANS). In soluble scruin, the two most reactive cysteines of soluble scruin are C837 and C900, whereas C146, C333, and C683 are moderately reactive. This pattern of reactivity is consistent with the topology of a typical beta-propeller protein; all of the reactive cysteines map to putative loops and turns whereas the unreactive cysteines lie within the predicted interior of the protein. The chemical reactivities of cysteine in the acrosomal process implicate C837 at an actin-binding site. In contrast to soluble scruin, in the acrosomal process, C837 is completely unreactive while the other cysteines become less reactive. Binding studies of chemically modified scruin correlate the extent of modification at C837 with the extent of inhibition of actin binding. Furthermore, peptides corresponding to residues flanking C837 bind actin and narrow a possible actin-binding region to a KQK sequence. On the basis of these studies, these results suggest that an actin-binding site lies in the C-terminal domain of scruin and involves a putative loop defined by C837 (Sun, 1997).

Limulus sperm contains a dynamic macromolecular structure that rapidly extends a 50 microm process called the true discharge. The core of this structure is a bundle of ordered filaments composed of a complex of actin, scruin and calmodulin. Its structure was determined by electron crystallographic reconstruction. The three-dimensional map reveals an actin-scruin helix that is azimuthally modulated by the influence of the interactions of a filament with its neighbors. There are a variety of density connections with neighboring filaments involving scruin. Scruin commonly contacts one neighbor, but up to three interfilament connections involving both domains of the 28 scruin molecules is observed in the unit cell. This structure indicates that promiscuous scruin-scruin contacts are the major determinants of bundle stability in the true discharge. It also suggests that rearrangements would be permitted, which can facilitate the transition from the coiled to the true discharge form (Sherman, 1999).

Other protein interactions of Kelch homologs

During activation of the Limulus sperm acrosomal process, actin filaments undergo a change in twist that is linked with the conversion from a coiled to a straight scruin-actin bundle. Since scruin had not been purified, its identity as an actin-binding protein has not been demonstrated. Using HECAMEG (methyl-6-O-(N-heptyl-carbamoyl)-alpha-D-glucopyranoside) detergent extraction in concert with high calcium, native scruin was purified and identified as an equimolar complex with calmodulin. 125I-Calmodulin overlays and calmodulin-Sepharose indicate that scruin binds calmodulin in calcium but not in EGTA. Overlay experiments also map the calmodulin binding site between the putative N- and C-terminal beta-propeller domains within residues 425-446. Immunofluorescence microscopy reveals that calmodulin colocalizes with scruin and actin in the coiled bundle. Although scruin binds calmodulin, pelleting assays and electron microscopy show that the scruin cross-links F-actin into bundles independent of calcium. Based on biochemical and structural studies, a model is suggested to explain how scruin controls a change in twist of actin filaments during the acrosome reaction. It is predicted that calcium subtly alters scruin conformation through its calmodulin subunit and the conformation change in scruin causes a shift in the relative positions of the scruin-bound actin subunits (Sanders, 1996).

Giant axonal neuropathy (GAN), an autosomal recessive disorder caused by mutations in GAN, is characterized cytopathologically by cytoskeletal abnormality. Based on its sequence, gigaxonin contains an NH2-terminal BTB domain followed by six kelch repeats, which are believed to be important for protein-protein interactions. A neuronal binding partner of gigaxonin is described. Results obtained from yeast two-hybrid screening, cotransfections, and coimmunoprecipitations demonstrate that gigaxonin binds directly to microtubule-associated protein (MAP)1B light chain (LC; MAP1B-LC), a protein involved in maintaining the integrity of cytoskeletal structures and promoting neuronal stability. Studies using double immunofluorescent microscopy and ultrastructural analysis reveal physiological colocalization of gigaxonin with MAP1B in neurons. Furthermore, in transfected cells the specific interaction of gigaxonin with MAP1B has been shown to enhance the microtubule stability required for axonal transport over long distance. At least two different mutations identified in GAN patients led to loss of gigaxonin-MAP1B-LC interaction. The devastating axonal degeneration and neuronal death found in GAN patients point to the importance of gigaxonin for neuronal survival. These findings may provide important insights into the pathogenesis of neurodegenerative disorders related to cytoskeletal abnormalities (Ding, 2002).

A subset of Kelch homologs binds ATP

Proteins with BTB/POZ domain are implicated in a broad variety of biological processes, including DNA binding, regulation of gene transcription and organization of macromolecular structures. Kelch domain containing BTB/POZ proteins like Mayven and Keap1 display limited sequence similarity with the actin-fragmin kinase from Physarum, a protein kinase with a kelch domain. Mouse Keap1, a Caenorhabditis elegans protein was named CKR, and human Mayven each bind 5'-p-fluorosulfonyl-benzoyl-adenosine (FSBA), a covalently modifying ATP analogue. Binding with 2-azido-ATP or ATP-Sepharose is also demonstrated. In contrast to Mayven, FSBA binding by CKR and Keap1 is specifically inhibited by excess ATP. The ATP binding pocket is located in the N-terminal half of Keap1. These findings indicate that several, but not all, BTB/POZ-kelch domain proteins possess an inconspicuous ATP binding cassette (T'Jampens, 2002).

kelch: Biological Overview | Regulation | Developmental Biology | Effects of Mutation | References

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.