Here, we isolated a novel F-actin–binding protein with a molecular mass of ~205 kD (p205). This protein was copurified with another protein with a molecular mass of 190 kD (p190) that lacked the F-actin–binding activity on various column chromatographies. The molecular cloning of the cDNAs of these two proteins revealed that the nucleotide sequence of the p190 cDNA was identical to that of the p205 cDNA, except for the two splicing regions. FISH analysis revealed that the genes of these two proteins were localized at the same locus. These results suggest that p205 and p190 are splicing variants derived from the same gene. Because a computer homology search revealed that the aa sequence of p190 was almost identical to that of human AF-6 protein, we theorize that p190 may be a rat counterpart of human AF-6 protein. We named p205 and p190 l- and s-afadins, respectively. Further purification steps of l-afadin, including Mono S column and Superdex 200 column chromatographies, did not separate l- and s-afadins from each other, and their elution profiles on these column chromatographies were similar but not identical. These proteins appeared at a position corresponding to a molecular mass of >600 kD on the gel filtration. These results suggest that 1- and s-afadins form a complex, but we cannot exclude the possibility that these two proteins do not form a complex, but instead are incidentally eluted at similar fractions on these column chromatographies. It is, therefore, possible that the complex is composed of a heterooligomer alone, a homooligomer alone, or the mixture. Western blot analysis indicated that the tissues other than brain expressed l-afadin alone. It is likely that l-afadin forms a homooligomer in these tissues.
Recently, Luna's group (Pestonjamasp et al., 1995
) has shown that a protein with a molecular mass of ~205 kD in bovine neutrophils has the 125
I-labeled F-actin–binding activity. When human neutrophils were subjected to SDS-PAGE, followed either by Western blot analysis using the antibody specific to l-afadin or by 125
I-labeled F-actin blot overlay, an immunoreactive band with a molecular mass of ~200 kD was detected, and this band showed the 125
I-labeled F-actin–binding activity (data not shown). This protein band is likely to be l-afadin, although its molecular mass was slightly smaller than that of rat l-afadin. The different molecular mass values may result from the species' differences. The exact relationship between the 205-kD protein described by Luna's group and l-afadin is not known, since the primary structure of the 205-kD protein has not been determined, but the 205-kD protein may be l-afadin.
l-Afadin showed 125
I-labeled F-actin–binding activity. Neither s-afadin nor the deletion mutant of l-afadin lacking the COOH-terminal 156 aa (myc–l-afadinΔC) showed this activity, whereas GST–l-afadin-C corresponding to the COOH-terminal 199 aa showed the 125
I-labeled F-actin– binding activity. The Mono Q sample of l-afadin increased the viscosity of F-actin on the low shear viscometry. Furthermore, His6–l-afadin-C corresponding to the COOH-terminal 199 aa was cosedimented with F-actin. These results indicate that l-afadin shows F-actin–binding activity and that the domain responsible for this activity is located at the COOH-terminal 199 aa. Many actin-binding protein families have been isolated and characterized. Of these families, for instance, the α-actinin/spectrin family members have been shown to usually form oligomers, such as a homodimer or a heterotetramer, and thereby show the F-actin–cross-linking activity. α-Actinin made the viscosity of F-actin unmeasurably high on the low shear viscometry, and it caused F-actin to associate into bundles and meshworks, as estimated by transmission electron microscopy. The stoichiometry is one α-actinin molecule per about seven actin molecules, and the kilodalton value is the order of 10 −7
(Meyer and Aebi, 1990
). In contrast, the effect of the Mono Q sample of l-afadin on F-actin was small on the low shear viscometry, and the sample hardly showed F-actin–cross-linking activity, as estimated by transmission electron microscopy. The stoichiometry was calculated to be 1 His6–l-afadin-C molecule per ~500 actin molecules. The kilodalton value was calculated to be the order of 10−7
. These results indicate that l-afadin belongs to a family different from the α-actinin/spectrin family. The ERM family, including ezrin, radixin, and moesin, belongs to the band 4.1 superfamily, an F-actin–binding protein superfamily (Bretscher, 1993
; Tsukita et al., 1997
), and there have been several reports that purified ezrin and moesin are hardly cosedimented with F-actin (Bretscher, 1983
; Pestonjamasp et al., 1995
; Shuster and Herman, 1995
). Luna's group (Pestonjamasp et al., 1995
) has recently shown by the 125
I-labeled F-actin blot overlay that ezrin and moesin exhibit the F-actin–binding activity. They have also shown by competition experiments using the 125
I-labeled F-actin blot overlay that ezrin and moesin bind along the sides of F-actin. The biochemical properties of l-afadin are apparently similar to those of ezrin and moesin. l-Afadin may bind F-actin in a manner similar to those of ezrin and moesin.
An immunofluorescence microscopic study by use of the antibody specific to l-afadin revealed that l-afadin was localized at the junctional complex region. Further microscopic and electron microscopic studies revealed that l-afadin was concentrated at cadherin-based cell-to-cell AJ. Taken together with the biochemical properties of l-afadin, l-afadin is likely to be involved in the linkage between the actin cytoskeleton and cell-to-cell AJ. Both l- and s-afadins have one PDZ domain. The PDZ domain has been found in a variety of proteins that are typically localized at cell-to-cell junctions (Saras and Heldin, 1996
) and shown to bind to the unique COOH-terminal motifs of target proteins, which are found in a large number of transmembrane proteins (Kim et al., 1995
; Kornau et al., 1995
; Niethammer et al., 1996
). l-Afadin may be recruited to cell-to-cell AJ through its PDZ domain. To understand the mechanism by which l-afadin is localized at cell-to-cell AJ, it is of crucial importance to identify its interacting molecule(s).
Many F-actin–binding proteins, including α-actinin, talin, and vinculin, have been shown to serve as linkers between the actin cytoskeleton and integrin at cell-to-matrix AJ (Jockusch et al., 1995
). The cytoplasmic domain of the β-subunit of integrin interacts directly with α-actinin and talin and indirectly with vinculin through α-actinin and talin (Jockusch et al., 1995
). At cell-to-cell AJ, the cytoplasmic domain of cadherin interacts with α-catenin through β-catenin (Ozawa et al., 1989
; Nagafuchi et al., 1991
; Takeichi, 1991
; Tsukita et al., 1992
). α-Catenin interacts directly with F-actin (Rimm et al., 1995
), α-actinin (Knudsen et al., 1995
), and ZO-1 (Itoh et al., 1997
). Vinculin is highly concentrated at cell-to-cell AJ, but its interacting molecule has not yet been identified (Geiger and Ginsberg, 1991
; Tsukita et al., 1992
). We showed here that l-afadin was localized at cell-to-cell AJ. Therefore, the molecular linkage mechanism between the actin cytoskeleton and cadherin is apparently similar to that between the actin cytoskeleton and integrin in the sense that F-actin is linked to each transmembrane protein through multiple F-actin–binding proteins.
gene has originally been identified as a fusion partner of the ALL-1
gene (Prasad et al., 1993
), which is known to be involved in acute leukemia (Cimino et al., 1991
). It has recently been shown that the AF-6
gene maps outside the minimal region of deletion in ovarian cancers with similar chromosomal aberrations (Saha et al., 1995
; Saito et al., 1996
). However, neither the significance of the fusion of these two genes nor the role of s-afadin in carcinogenesis has been clarified. s-Afadin has recently been shown to have significant homology to the product of the Drosophila canoe
gene (Miyamato et al., 1995). The canoe
gene interacts genetically with the Notch
signal cascade in the eye, bristle, and wing development (Miyamato et al., 1995). Moreover, the canoe
gene interacts genetically with the armadillo
) gene (Miyamato et al., 1995). The products of the Notch
genes have been implicated in adhesive cell-to-cell communications (Peifer et al., 1993
; Artavanis-Tsakonas et al., 1995
). l- and s-Afadins may also play a role in the development of various tissues by mediating interactions between the adhesive molecule–initiated signal transduction pathways.
The AF-6 protein (s-afadin) has recently been shown to interact specifically with the GTP-bound active form of Ras small G protein (Kuriyama et al., 1996
); however, we have not been able to reproduce this earlier finding. We found that both the GDP- and GTP-bound forms of Ki-Ras interacted with l- and s-afadins to similar extents and that the stoichiometry of these interactions was negligible (data not shown). Moreover, an immunofluorescence microscopic study of Madin-Darby canine kidney cells stably expressing an myc-tagged dominant active mutant of Ki-Ras (N12V) or an myc-tagged wild type of Ki-Ras showed that these two types of Ras are not concentrated at cell-to-cell AJ (Takaishi, K., and Y. Takai, unpublished observations). All of these results do not support the specific interaction of Ras with s-afadin, and the relationship between Ras and s-afadin is currently unknown.