Danio rerio αE-catenin Is a Monomeric F-actin Binding Protein with Distinct Properties from Mus musculus αE-catenin*

Background: Mammalian αE-catenin is an allosterically regulated F-actin binding protein that inhibits the Arp2/3 complex. Results: D. rerio αE-catenin is a monomer that binds β-catenin and F-actin simultaneously but does not inhibit the Arp2/3 complex. Conclusion: D. rerio αE-catenin directly links the cadherin-catenin complex to F-actin. Significance: Core functions but not regulatory properties are conserved between α-catenin orthologs. It is unknown whether homologs of the cadherin·catenin complex have conserved structures and functions across the Metazoa. Mammalian αE-catenin is an allosterically regulated actin-binding protein that binds the cadherin·β-catenin complex as a monomer and whose dimerization potentiates F-actin association. We tested whether these functional properties are conserved in another vertebrate, the zebrafish Danio rerio. Here we show, despite 90% sequence identity, that Danio rerio and Mus musculus αE-catenin have striking functional differences. We demonstrate that D. rerio αE-catenin is monomeric by size exclusion chromatography, native PAGE, and small angle x-ray scattering. D. rerio αE-catenin binds F-actin in cosedimentation assays as a monomer and as an α/β-catenin heterodimer complex. D. rerio αE-catenin also bundles F-actin, as shown by negative stained transmission electron microscopy, and does not inhibit Arp2/3 complex-mediated actin nucleation in bulk polymerization assays. Thus, core properties of α-catenin function, F-actin and β-catenin binding, are conserved between mouse and zebrafish. We speculate that unique regulatory properties have evolved to match specific developmental requirements.

The cadherin-catenin complex mediates cell-cell adhesion (1) and is essential for normal development and tissue organization in the Metazoa (2). ␣E-catenin, a central component of the cadherin-catenin complex, is conserved throughout the Metazoa. In mammals, ␣E-catenin functionally links the cadherin⅐␤-catenin complex to the actin cytoskeleton and regulates cell-cell adhesion and cell migration (2)(3)(4)(5)(6)(7)(8). This link is dynamic or regulated because mammalian ␣E-catenin binds the cadherin⅐␤-catenin complex as a monomer and F-actin as a homodimer in bulk assays (9 -13). Mammalian ␣E-catenin also stabilizes cell-cell contacts and inhibits cell migration by organizing F-actin (9) and inhibiting Arp2/3 complex-mediated nucleation of F-actin (12,14).
Although the functions of mammalian ␣E-catenin are well defined, it is unknown whether its structural and biochemical properties are conserved in other organisms (15,16). Previous studies examined whether the biochemical properties of mammalian ␣E-catenin are conserved in orthologs from Caenorhabditis elegans (17,18) and Dictyostelium discoideum (19). The amino acid sequence identities of C. elegans HMP-1 (␣-catenin) and D. discoideum ␣-catenin with M. musculus ␣E-catenin are low, 37 and 14%, respectively, and regulatory properties are not well conserved (12,(17)(18)(19): C. elegans HMP-1 is an autoinhibited monomer that binds weakly to F-actin in solution (17,18), and D. discoideum ␣-catenin is a monomer that constitutively binds and bundles F-actin but does not inhibit Arp2/3-mediated actin nucleation (19). It is perhaps not surprising that these distantly related orthologs of ␣-catenin exhibit different organizational and functional properties. Therefore, we examined a closely related ortholog in the zebrafish, D. rerio, which is 90% identical in amino acid sequence to M. musculus ␣E-catenin. Despite the considerable sequence identity, there are remarkable differences in the functional properties of D. rerio ␣E-catenin.

EXPERIMENTAL PROCEDURES
Recombinant Protein Expression and Purification-DNA encoding full-length (amino acids 1-907) D. rerio ␣E-catenin was inserted into an Smt3 (Saccharomyces cerevisiae SUMO) adapted pET28 (20) vector to generate an in-frame fusion between D. rerio ␣E-catenin and N-terminal His 6 -Smt3 tag. Recombinant fusion protein was expressed in BL21(DE3) Codon Plus E. coli cells (11), purified on nickel-nitrilotriacetic acid-agarose beads, and eluted with imidazole. Protein was applied to a MonoQ anion exchange column and eluted at 200 mM NaCl in 20 mM Tris, pH 8.0, 1 mM DTT with a 0 -1 M NaCl gradient. The His-Smt3 tag was removed by overnight incubation at 4°C with Smt3 protease (Ulp1) at a 1:1000 Ulp1:D. rerio ␣E-catenin mass ratio. D. rerio ␣E-catenin was further purified by Superdex 200 gel filtration chromatography in 20 mM Tris, pH 8.0, 150 mM NaCl, 10% glycerol and 1 mM DTT. Pure protein, eluted as a monomer at a concentration of 40 -50 M, was flash frozen in liquid nitrogen and stored at Ϫ80°C. His 6 -TEV-D. rerio ␤-catenin (21) was expressed in BL21(DE3) Codon Plus E. coli cells, purified on nickel-nitrilotriacetic acid-agarose beads, and cleaved off the beads by overnight incubation with TEV protease at 4°C (where TEV is nuclear inclusion a (NIa) protease encoded by the tobacco etch virus). D. rerio ␤-catenin was purified on a MonoQ anion exchange column followed by a Superdex 200 gel filtration column in the same conditions as described for D. rerio ␣E-catenin. Recombinant full-length M. musculus ␣E-catenin (1-906) and M. musculus ␤-catenin (1-781) were expressed and purified as reported previously (13).
Small Angle X-ray Scattering-Small angle scattering data were measured at beamline 4-2 at the Stanford Synchroton Radiation Laboratory. D. rerio ␣E-catenin was prepared in 20 mM Tris, pH 8.0, 150 mM NaCl, 1 mM DTT, and 1% glycerol. Samples at concentrations of 5, 10, 20, and 30 M were loaded into a 1.5-mm quartz capillary flow cell maintained at 20°C, and 10 ϫ 5-s exposures were measured from each concentration. The raw scattering data were normalized to the incident beam intensity, and buffer scattering was subtracted. The radius of gyration (R g ) 3 was computed using the program Primus (22). Data used for estimating R g were restricted to scattering angles for which the product q ϫ (estimated R g ) Յ 1.3 (q ϭ 4sin()/). The molecular mass of D. rerio ␣E-catenin in solution was obtained from I(0) by extrapolation to q ϭ 0 using water as the calibration standard and assuming a protein partial specific volume of 0.7586 cm 3 /g (23).
Limited Proteolysis and Edman Sequencing-12 M D. rerio ␣E-catenin and 12 M M. musculus ␣E-catenin were incubated at room temperature in 0.05 mg/ml sequence-grade trypsin (Roche Applied Science) in 20 mM Tris, pH 8.0, 150 mM NaCl, and 1 mM DTT. Reactions were stopped with 2ϫ Laemmli buffer at the indicated times, and samples were analyzed by SDS-PAGE. For N-terminal sequencing, digested peptides were blotted onto PVDF membrane, stained with 0.1% Coomassie (R-250/40%methanol/1% acetic acid), destained, and dried. Individual bands were excised and sequenced by Edman degradation.
Isothermal Titration Calorimetry-ITC experiments were performed on a VP-ITC calorimeter (Microcal, GE Healthcare). For D. rerio ␣E-catenin/D. rerio ␤-catenin binding, a total of 31 9-l aliquots of 100 M D. rerio ␣E-catenin were injected into the cell, which contained 10 M D. rerio ␤-catenin in 20 mM Hepes, pH 8.0, 150 mM NaCl, and 1 mM DTT at 25°C. For D. rerio ␣E-catenin/M. musculus ␤-catenin binding, a total of 31 9-l aliquots of 100 M M. musculus ␤-catenin were injected into the cell, which contained 10 M D. rerio ␣E-catenin in 20 mM Hepes, pH 8.0, 150 mM NaCl, and 1 mM DTT at 25°C. Heat change was measured for 240 s between injections. For each experiment, the average value calculated from heat changes measured at saturation was subtracted from all data points.
Actin Cosedimentation Assays-Chicken muscle G-actin was incubated in 1ϫ actin polymerization buffer (20 mM Tris, pH 8.0, 100 mM KCl, 2 mM MgCl 2 , 0.5 mM ATP, and 1 mM EGTA) for 1 h at room temperature to polymerize filaments. Gel filtered D. rerio ␣E-catenin or D. rerio ␣E-catenin and D. rerio⅐M. musculus ␤-catenin heterocomplex was diluted to indicated concentrations in 1ϫ reaction buffer (20 mM Tris, pH 8.0, 150 mM NaCl, 2 mM MgCl 2 , 0.5 mM ATP, 1 mM EGTA, and 1 mM DTT) with and without 2 M F-actin and incubated for 30 min at room temperature. Samples were centrifuged at 100,000 rpm for 20 min in a TLA 120.1 rotor. Pellets were resuspended in 1ϫ Laemmli sample buffer, separated by SDS-PAGE, and stained with Coomassie Blue. Gels were imaged on a LI-COR scanner and measured and quantified in ImageJ software. Binding data were processed with Prism software.
Transmission Electron Microscopy-Protein samples were prepared as for actin cosedimentation assays and deposited on carbon grids. Samples were stained with 1% uranyl acetate for 3 min and examined in a JEOL TEM1230 electron microscope at 12,000ϫ magnification. Brightness and contrast were adjusted in ImageJ software. No other manipulations were performed.
Actin Polymerization Assays-10% pyrene-labeled rabbit muscle G-actin (Cytoskeleton, Inc.) was diluted into fresh G buffer (20 mM Tris, pH 8.0, 0.2 mM CaCl 2 , 0.2 mM ATP, and 1 mM DTT) immediately before use. 10% Pyrene-labeled G-actin was centrifuged for 30 min at 14,000 rpm in FA 45-30-11 rotor (Eppendorf) to remove aggregates. Prior to each experiment, Ca 2ϩ was exchanged for Mg 2ϩ by adding 1 ⁄ 10 volume of 10 mM EGTA and 1 mM MgCl 2 and incubating 2 min at room temperature. 10ϫ Polymerization buffer was then added to a final concentration of 1ϫ (20 mM Tris, pH 8.0, 100 mM KCl, 2 mM MgCl 2 , 0.5 mM ATP, and 1 mM EGTA) to initiate polymerization. Pyrene fluorescence (365 nm excitation, 407 nm emission) was measured every 10 s until the fluorescence reached a stable value (1000 s). Measurements were taken using a Tecan Infinite M1000 plate reader. The delay between addition of polymerization buffer and commencement of measurements was 10 -20 s for all experiments. Where appropriate, bovine Arp2/3 com-plex and human GST-WASp-VCA were added to the reactions with polymerization buffer to a final concentration of 50 nM each. Full-length D. rerio ␣E-catenin, M. musculus ␣E-catenin, or rabbit ␣-actinin were gel filtered into 1ϫ polymerization buffer prior to the experiment and added along with the polymerization buffer to a final concentration of 1-8 M.
Thus, D. rerio ␣E-catenin is a member of the ␣-catenin family and is closely related to M. musculus ␣E-catenin.
To determine whether the domain organization of D. rerio ␣E-catenin in solution is similar to M. musculus ␣E-catenin, we performed limited trypsin proteolysis (11). After 120 min of digestion, two stable fragments of D. rerio ␣E-catenin were detected by SDS-PAGE (Fig. 1C). N-terminal sequencing of these two fragments revealed they started at residues 82 and 385 respectively, which correspond to the proteolytically resistant domains identified previously in M. musculus ␣E-catenin (11,24), the dimerization/␤-catenin-binding domain (residues 82-287) and the M-domain (residues 385-651) (Fig. 1C). Despite conservation of potential cleavage sites in the sequences, a protease-resistant dimerization/␤-catenin-binding domain is not observed in C. elegans HMP-1 (17) or D. discoideum ␣-catenin (19). This suggests reduced proteolytic sensitivity of N-terminal domains in vertebrate ␣E-catenins. A protected fragment of the actin-binding domain was not identified in either D. rerio ␣E-catenin or M. musculus ␣E-catenin. These results, taken together with sequence analysis, demon-   strate that D. rerio ␣E-catenin is closely related to M. musculus ␣E-catenin.
D. rerio ␣E-catenin Is a Monomer-We examined the oligomerization state of recombinant full-length D. rerio ␣E-catenin by size exclusion chromatography using Superdex 200, native PAGE, and small angle x-ray scattering (SAXS) (17,19). D. rerio ␣E-catenin eluted as a discrete peak with an estimated molecular weight of 100 kDa by size exclusion chromatography ( Fig.  2A), and migrated as a single band on native PAGE (Fig. 2B). SAXS curves were measured for D. rerio ␣E-catenin at a range of concentrations from 5-30 M in solution (Fig. 2C). The predicted molecular mass of D. rerio ␣E-catenin monomer by sequence analysis is 100.5 kDa. The molecular mass for D. rerio ␣E-catenin obtained from the SAXS data is 100.3 Ϯ 1.5 kDa, strongly supporting the conclusion that D. rerio ␣E-catenin is a monomer even at high concentrations. Furthermore, SAXS analysis yielded a radius of gyration (R g ) for D. rerio ␣E-catenin of 43.9 Å (Fig. 2C), which is similar to that of the M. musculus ␤-␣E-catenin chimera monomer (R g ϭ 44.6 Å; Fig. 2D; 17). The ␤-␣E-catenin chimera is a ␤-catenin-stabilized M. musculus ␣E-catenin monomer used because the contaminating presence of M. musculus ␣E-catenin dimer (R g ϭ 59.6 Å; 17) in preparations of M. musculus ␣E-catenin monomer (11) prevents accurate SAXS measurements. The R g values indicate that D. rerio ␣E-catenin has an overall conformation similar to monomeric M. musculus ␣E-catenin but is less compact than the auto-inhibited monomeric proteins C. elegans HMP-1 and M. musculus vinculin (R g ϭ 37.2 and 36.7 Å, respectively; Fig.  2D) (17,25). Unlike mammalian ␣E-catenin, but similar to C. elegans HMP-1 (17) and D. discoideum ␣-catenin (R g ϭ 32.8 Å; Fig. 2D) (19), D. rerio ␣E-catenin is a monomer in solution, at least to the highest measured concentration of 30 M, well above the estimated cytosolic concentration of ␣E-catenin in mammalian epithelial cells (0.6 M) (12).
To investigate whether D. rerio ␣E-catenin bundles F-actin, we used transmission electron microscopy to visualize negative stained F-actin incubated with D. rerio ␣E-catenin (Fig. 4C) (9,19). D. rerio ␣E-catenin bundled 2 M F-actin at 8 M but not at lower concentrations. In contrast, M. musculus ␣E-catenin bundled 2 M F-actin at 1-2 M (Fig. 4C). Activation of a cryptic D. rerio ␣E-catenin dimerization domain upon actin binding or the presence of a second F-actin binding site with lower affinity on the N terminus could be responsible for F-actin weak bundling. Similar properties have been reported in ␣-catenin homologs: vinculin tail dimerizes only upon binding to F-actin (27), and the N-terminal domains of both D. discoideum ␣-catenin (19) and mammalian ␣E-catenin (9) have been shown to bind F-actin in vitro. Further work is required to determine the mechanism by which D. rerio ␣E-catenin interacts with actin filaments.
A Heterodimer of D. rerio ␣E-catenin and ␤-Catenin Binds F-actin-We performed high-speed cosedimentation assays of purified D. rerio ␣/␤-catenin heterodimer with F-actin to investigate whether ␤-catenin regulates the binding of D. rerio ␣E-catenin to F-actin (9,11). The D. rerio ␣E-catenin⅐D. rerio ␤-catenin heterodimer complex cosedimented with F-actin above background (Fig. 5A). At concentrations higher than 6 -8 M, the D. rerio ␣E-catenin⅐D. rerio ␤-catenin heterodimer sedimented independently of actin such that we were unable to saturate binding (Fig. 5B). Increasing the NaCl concentration in the reaction buffer to 300 mM and/or performing the experiment at 4°C to limit aggregation did not reduce background sedimentation. Although saturation was not achieved, we can fit the curve with a single site-binding model to give a lower boundary on the K d value (9.9 Ϯ 4.5 M; Fig. 5B). Thus, the D. rerio ␣⅐␤-catenin heterodimer binds F-actin at least 10ϫ more weakly than the D. rerio ␣E-catenin monomer (Fig. 4). Interestingly, the heterodimer of D. rerio ␣E-catenin and M. musculus ␤-catenin also cosedimented with F-actin above background but with at least an order of magnitude weaker affinity than the D. rerio ␣E-catenin monomer (Fig. 5, C and D).

Correlating Biochemical Properties of D. rerio ␣E-catenin and M. musculus ␣E-catenin with Loss of Function Phenotypes-
The biochemical data presented here show that D. rerio ␣E-catenin binds to F-actin by itself or in the presence of ␤-catenin, unlike orthologs in M. musculus (12)(13)26) and C. elegans (17). Thus, D. rerio ␣E-catenin may provide a direct link between the cadherin⅐␤-catenin complex and the actin cytoskeleton. However, this activity of D. rerio ␣E-catenin is regulated: the affinity of the ␣E-catenin⅐␤-catenin complex for F-actin is at least an order of magnitude weaker than that of D. rerio ␣E-catenin alone. Also, unlike mammalian ␣E-catenin, D. rerio ␣E-catenin does not inhibit Arp2/3 complex-mediated actin nucleation. Do these differences in biochemical properties of M. musculus ␣E-catenin and D. rerio ␣E-catenin reflect differences in loss-of-function phenotypes in the mouse and zebrafish, respectively? Loss of ␣E-catenin function in mammals (3)(4)(5)(6)(7) and zebrafish (34) leads to a decrease in cadherin-mediated cell-cell adhesion. However, in mammals, loss of ␣E-catenin increases Arp2/3 complex-dependent lamellipodial dynamics (5,14), whereas in zebrafish, D. rerio ␣E-catenin depletion increases membrane blebbing (34). Membrane blebbing is caused by localized loss of attachment of the plasma membrane to the cortical cytoskeleton, which allows hydrostatic pressure in the cytoplasm to induce membrane protrusions independent of the Arp2/3 complex (35). The ability of D. rerio ␣E-catenin to directly link ␤-catenin to F-actin in vitro suggests the cadherin⅐catenin complex may function as a membrane anchor for the cortical actomyosin cytoskeleton in zebrafish, as proposed for mammalian cells (36) and D. melanogaster (37).
Disruption of adhesion tension in gastrulation stage zebrafish embryos (38) supports the role of the zebrafish cadherin⅐catenin complex as a membrane anchor for the cortical actin cytoskeleton. Physical separation of ectodermal progenitor cells disrupts adhesion tension, causing distortions in plasma membrane curvature, such that D. rerio ␣E-catenin and F-actin no longer co-localize with ␤-catenin and E-cadherin at the plasma membrane (38). However, D. rerio ␣E-catenin and F-actin co-localize with one another in the cytoplasm (38). Despite clear differences in the biochemical activities of M. musculus and D. rerio ␣E-catenin, further work is required to understand the relationship between these in vitro activities with in vivo functional differences.
D. rerio ␣E-catenin and Protein Evolution Studies-We find functional differences between two vertebrate ␣E-catenins, mouse and zebrafish, that are 90% identical in sequence. Our results suggest that changes in a small number of amino acids have a critical role in the evolution of structure and function in ␣E-catenin. 26 of 907 residues are not similar in size and/or charge between D. rerio and M. musculus ␣E-catenin. We posit that adaptive changes in regions that regulate ␣-catenin conformation caused mammalian and zebrafish ␣E-catenins to develop distinct functional properties to match specific developmental requirements. However, further work is required to identify the specific residues that regulate function in vertebrate ␣E-catenin orthologs. Moreover, every ␣-catenin heretofore characterized has exhibited distinct biochemical properties (12,(17)(18)(19), and no cohesive model of ␣-catenin structure/ function evolution exists. The evolutionary development of ␣-catenin as a direct link between ␤-catenin at the adherens junction and the actin cytoskeleton is unresolved. Although data presented here suggest D. rerio ␣E-catenin can act as a direct link, work with orthologs from mammals (12-13) and C. elegans (17) indicate that the linkage is indirect and/or tightly regulated. We propose that association of ␣-catenin with ␤-catenin is a conserved mechanism for negative regulation of actin binding. It remains to be determined whether homodimerization, strong F-actin bundling, and Arp2/3 inhibition are required properties of ␣-catenins in other metazoans   or instead specific characteristics of mammalian ␣E-catenin. Examining orthologs from several other organisms and investigating directly the effects of amino acid changes will provide information to develop a model for ␣E-catenin structure/function evolution.