March 2007
Volume 48, Issue 3
Free
Retinal Cell Biology  |   March 2007
Retina-Specific Protein Fascin 2 Is an Actin Cross-linker Associated with Actin Bundles in Photoreceptor Inner Segments and Calycal Processes
Author Affiliations
  • Jennifer Lin-Jones
    From the Department of Molecular and Cell Biology, University of California at Berkeley, Berkeley, California.
  • Beth Burnside
    From the Department of Molecular and Cell Biology, University of California at Berkeley, Berkeley, California.
Investigative Ophthalmology & Visual Science March 2007, Vol.48, 1380-1388. doi:10.1167/iovs.06-0763
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Jennifer Lin-Jones, Beth Burnside; Retina-Specific Protein Fascin 2 Is an Actin Cross-linker Associated with Actin Bundles in Photoreceptor Inner Segments and Calycal Processes. Invest. Ophthalmol. Vis. Sci. 2007;48(3):1380-1388. doi: 10.1167/iovs.06-0763.

      Download citation file:


      © 2016 Association for Research in Vision and Ophthalmology.

      ×
  • Supplements
Abstract

purpose. Fascin 2 is a retinal-specific member of the fascin family of actin filament-bundling proteins. Fascin 2 mutation in humans results in autosomal dominant retinitis pigmentosa or macular degeneration. To investigate the role of fascin 2 in photoreceptor survival, the authors examined its localization in photoreceptors and characterized its interactions with actin filaments in vitro.

methods. Fascin 2 localization was determined by immunohistochemistry and transgenic expression of green fluorescent protein (GFP)–tagged fascin 2 in Xenopus laevis rods. Fascin 2 actin-binding and actin-bundling activity were examined in sedimentation assays using bacterially expressed fusion proteins and polymerized actin. To assess the role of phosphorylation of a conserved serine (amino acid 39) in fascin 2 on subcellular localization and actin-binding, effects of serine mutants were also examined in transgenic Xenopus and in in vitro assays.

results. Fascin 2 is localized to actin filament bundles of the photoreceptor inner segment and calycal processes. Like fascin 1, fascin 2 binds and cross-links actin filaments. Mutation of serine 39 to an aspartic acid reduced fascin 2 binding of actin filaments and abolished fascin 2 bundling of actin filaments in vitro but produced no detectable effect on GFP-tagged fascin 2 localization in transgenic Xenopus.

conclusions. These observations suggest that fascin 2 plays a role in the assembly or stabilization of inner segment and calycal process actin filament bundles in photoreceptors and that serine 39 phosphorylation reduces actin-binding and cross-linking activity and, thus, is likely to regulate the inner segment actin cytoskeleton.

Fascin 2 is a retinal-specific member of the fascin family of actin filament-bundling proteins. 1 Of the three known fascin family members, fascin 1 has been studied most extensively; fascin 3 is expressed exclusively in the testis. 2 Fascin 1, found in vertebrates and invertebrates, is located in bundled actin structures such as filopodia, microspikes, and stress fibers in vivo and in cultured cells. 3 Fascin 1 is a globular, monomeric protein that cross-links actin filaments using two actin-binding sites, one near the N-terminus and the other in the C-terminal half. 4 5 The precise domains responsible for actin filament binding have not been mapped for any of the fascins. However, a serine residue (amino acid 39), found in all but one of the fascins, is contained within a conserved domain with sequence homology to the actin-binding site of myristoylated alanine-rich C-kinase substrate (MARCKS). Serine 39 of fascin 1 is the major phosphorylation target site for protein kinase C, and serine 39 phosphorylation has been shown to regulate fascin 1 binding to actin filaments. 5 6 The phosphorylation state of serine 39 has also been shown to influence cell migration on different extracellular matrices. 5 7 Although fascin 1 interactions with actin filaments have been well characterized, recent reports have shown that fascin 2 also binds actin filaments and organizes them into stiff bundles. 8  
It is clear that fascin 2 plays an important role in human photoreceptor survival. A single frameshift mutation in the fascin 2 gene, which results in premature termination of protein translation, is responsible for autosomal dominant retinitis pigmentosa or autosomal dominant macular degeneration in different families. 9 10 A phenotype similar to that of retinitis pigmentosa is also produced in mice heterozygous for the same frameshift mutation in the fascin 2 gene and in mice with a disrupted fascin 2 gene. 11 Heterozygous and homozygous mice with a mutated or disrupted fascin 2 gene exhibit progressive photoreceptor loss, depression of ERGs, and outer segment shortening. Degeneration in homozygotes is more rapid and severe than in heterozygotes. Nonetheless, normal-looking photoreceptors develop even in homozygous mice. Thus, fascin 2 does not appear to be required for photoreceptor morphogenesis, but decreased fascin 2 levels contribute to progressive photoreceptor degeneration. 
The actin cytoskeleton of photoreceptors has three distinct components. In photoreceptor inner segments, arrays of parallel actin bundles originate with their plus-ends at the tips of microvilli-like calycal processes that encompass the base of the outer segment and extend through the ellipsoid between the plasma membrane and the ellipsoid mitochondria. 12 Actin filaments are also located in the distal connecting cilium near the proximal outer segment, where disk morphogenesis occurs, and in photoreceptor synapses. 13 Because the disruption of the photoreceptor actin cytoskeleton with cytochalasin results in abnormal disk morphogenesis, it has been proposed that the connecting cilium subpopulation of actin filaments is required for proper disk formation. 14  
Fascin 2 was localized to the inner and outer segments of the photoreceptor layer of bovine retinas by immunohistochemistry. 8 Based on this light level localization, the authors hypothesized that fascin 2 might interact with the actin filament population in the distal connecting cilium associated with disk morphogenesis. However, the precise localization of fascin 2 relative to the photoreceptor actin cytoskeleton has not been examined at higher resolution. 
To better understand the function of fascin 2 in photoreceptors, we analyzed fascin 2 subcellular localization in photoreceptors of fish and Xenopus laevis, in which inner segment actin bundles are more prominent and thus easier to resolve than in mammals. We have also shown that like fascin 1, fascin 2 actin filament binding, and fascin 2-mediated bundling in vitro is affected by the phosphorylation state of the highly conserved serine 39 residue. 
Materials and Methods
Xenopus and Zebrafish Fascin 2 Cloning
The entire coding sequence for one D. rerio (zebrafish) fascin 2 (DrF2A) was found in a search of the NCBI GenBank and EMBL databases. The search also found partial sequences for a X. laevis fascin 2 (XF2) and a second D. rerio fascin 2 (DrF2B). The remaining coding sequences for XF2 and DrF2B were amplified from Xenopus (provided by Joe Besharse at the Medical College of Wisconsin) and zebrafish (a gift from Susan Brockerhoff, University of Washington) retinal libraries with gene-specific and T7 promoter primers and cloned into PCR 2.1 (Invitrogen, Carlsbad, CA) for sequencing. PCR clones using gene-specific primers with added restriction sites for cloning were also amplified from these libraries to generate plasmids containing the entire XF2, DrF2A, and DrF2B coding sequences. The primers E-MSTNG (GATTCATGTCTACAAACGGAATAAGCGCAGC) and SSLWEH-X (CTCATCCTTGTGGGAGCACTGAGCTCGAC) were used to amplify DrF2A. E-ARMP (GAATTCATGCCCTCCAATGGCACCAAAGC) and X-H*YE (CTCTTCCCTCTGGGAATACTGACATC) primers were used to amplify DrF2B, and BH-KENM (GGATCCAAGCTTAAGGAAAATATGCCAGCCAATGG) and X-SCTC (GGGAGTACTAATGCACATGCAGCACCTCGAC) primers were used to amplify XF2. The coding sequences for the Xenopus and 2 zebrafish fascin 2 genes were subcloned into XOP4/GFP, a vector with a green fluorescent protein (GFP) tag driven by a rod-specific promoter for Xenopus transgenesis. 15 The three fascin 2 sequences were also cloned into pGEX4T-1 (GE Healthcare, Piscataway, NJ), and pMAL (New England Biolabs, Beverly, MA) expression vectors for protein purification. Mutations of XF2 serine residue 39 to either an alanine (S39A) or an aspartic acid (S39D) were created by PCR using the KVYASAPAL (TAAGTGTATGCTAGCGCCCCAGCCCTG) primer for the S39A mutant and the KVYASAPDL (TAAAGTGTATGCTAGCGCCCCAGACCTG) primer for the S39D mutant. Mutant sequences were cloned into XOP4/GFP/XF2 for transgenic analysis. Serine 39 in the DrF2B sequence was also mutated to aspartic acid by amplification with the WIQKKKLD (CCAGATCTGCTTCTTCTTGAGGTC) primer and cloned into the glutathione S-transferase (GST)/DrF2B fusion plasmid for actin-binding and bundling assays. 
Generation of Fascin 2 Antibodies
Maltose-binding protein (MBP) fusions to XF2, DrF2A, and DrF2B were expressed in BL21 cells and purified with amylose resin. A mixture of the three purified proteins was injected into rabbits to produce polyclonal antisera (Bethyl Laboratories, Montgomery, TX). The resultant antiserum was purified by affinity chromatography with maltose-binding protein (MBP) fusion to the three fascin 2 proteins, and antibodies against the MBP tag were removed with an MBP column. 
Xenopus Transgenesis
Linearized plasmid DNA containing GFP sequence fused to XF2, XF2S39D, XF2S39A, DrF2A, or DrF2B was used to inject into X. laevis eggs to generate transgenic tadpoles expressing protein under the control of a modified Xenopus rod opsin promoter (XOP4), as has been described elsewhere. 15 Normal surviving tadpoles were screened for green eye fluorescence after 3 to 4 days. Ten-day-old transgenic and wild-type tadpole eyes were fixed, cryosectioned, and incubated with Texas-Red phalloidin. 15 More than seven tadpole retinas were examined for each of the different transgenes. Sections were imaged with confocal microscopy at the Molecular Imaging Center at the University of California at Berkeley or by fluorescence microscopy and deconvolution (Openlab software; Improvision Inc., Lexington, MA). 16 All animals used in this study were handled according to the guidelines outlined in the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 
Western Blot Analysis and Immunohistochemistry
Eyes from adult zebrafish were dissected, homogenized in phosphate-buffered saline containing 1 mM dithiothreitol and a mixture of protease inhibitors, 17 and prepared for SDS-PAGE. Shakeoffs (isolated photoreceptor inner/outer segments) were prepared from green sunfish (Lepomis cyanellus) retinas as previously described 17 and collected by a low-speed (4000g) microfuge spin at 4°C. The shakeoff pellet was resuspended in 1× SDS-PAGE sample buffer. Protein extracts were electrophoresed on 4% to 12% Bis-Tris NuPAGE gels (Invitrogen) and transferred to an Immobilon-P membrane (Millipore, Bedford, MA). A 1:10,000 dilution of the fascin 2 antibody was used, and the blot was processed as described previously. 17  
Adult zebrafish eyes were fixed overnight at 4°C (Cytoskelfix; R and D Enterprises, Lakewood, CO) and processed for cryosectioning, as previously described. 15 Ten-micrometer sections were immunostained with a 1:2000 dilution of the fascin 2 antibody and a 1:400 dilution of an actin monoclonal antibody (MP Biomedicals, Aurora, OH) and were visualized with the use of Cy-3 goat anti–rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) and Alexa Fluor 488 goat anti–mouse IgG (Invitrogen) secondary antibodies. Photoreceptor inner/outer segment preparations from sunfish retinal shakeoffs were stained with the fascin 2 antibody, Texas Red-phalloidin (Invitrogen), and an acetylated tubulin monoclonal antibody (Sigma-Aldrich, St. Louis, MO), as previously described. 17 Images of fluorescence-labeled retinal sections and shakeoffs were acquired on the confocal microscope at the University of California at Berkeley Molecular Imaging Facility. 
Actin-Binding and -Bundling Assays
Bacterially expressed GST fusion proteins were purified and prespun at 150,000g for 1 hour at 24°C to remove any precipitated protein. Varying concentrations of fusion proteins were incubated with 7.1 μM nonmuscle actin filaments (Cytoskeleton, Inc., Boulder, CO) in 5 mM Tris-HCl, pH 8.0, 0.2 mM CaCl2, 50 mM KCl, 2 mM MgCl2, and 1 mM ATP for 30 minutes at 24°C. Actin-binding samples were spun at 150,000g for 40 minutes at 24°C. Actin-bundling samples were spun at 10,000g for 20 minutes at 24°C. To visualize actin bundles, bundling reactions were incubated on poly-lysine–coated coverslips for 30 minutes at 24°C, fixed for 30 minutes on ice with 3% formaldehyde/PBS, stained with a 1:50 dilution of Alexa Fluor 488 phalloidin (Invitrogen) in PBS, and visualized by confocal laser microscopy. For sedimentation assays, equal amounts of the supernatants and pellets in 1× SDS-PAGE sample buffer were analyzed by SDS-PAGE. Densitometry of SDS-PAGE gels was performed using NIH Image. 
Results
Cloning Zebrafish and Xenopus Fascin 2
Fascin 2 genes were identified from zebrafish and from X. laevis by using bovine and human fascin 2 sequences to search the NCBI and EMBL sequence databases. A partial sequence containing the C-terminal of fascin 2 was found for Xenopus. Two different fascin 2 sequences were found for D. rerio; one encompassed the entire coding sequence and the other encoded only the N-terminal portion. Using gene-specific and plasmid vector primers, we cloned the entire coding regions of all three of these fascin 2 genes from retinal cDNA libraries (kindly provided by Joe Besharse at the Medical College of Wisconsin and Susan Brockerhoff at the University of Washington). Figure 1is a ClustalW alignment of fascin 2 sequences comparing the newly cloned X. laevis and zebrafish fascin 2 sequences to those of human, bovine, mouse, rat, Xenopus tropicalis, and Fugu rubripes from the sequence database. 
Fascin 2 genes are remarkably conserved throughout the entire coding sequence among amphibians, mammals, and fish. The two zebrafish fascin 2 genes are 77% identical (86% similar), and both have equivalent sequence identity (61%–63%) and similarity (77%–78%) to human fascin. 2 We have arbitrarily designated the full-length zebrafish fascin 2 sequence already existing in the sequence database as DrF2A and the second zebrafish fascin 2 as DrF2B. X. laevis fascin 2 (XlF2) also exhibits a remarkable degree of similarity (81%) to human fascin 2. The putative actin-binding domain near the N-terminus (the MARCKS homology domain) is conserved in all fascin 2 genes, as it is in all other members of the fascin family. 
Fascin 2 Colocalizes with Photoreceptor Inner Segment Actin Bundles
To examine the subcellular localization of fascin 2 in frog and fish retina, antibodies were made against a mixture containing Xenopus and each zebrafish fascin 2 expressed as fusion proteins in bacteria and purified by affinity chromatography. In Western blots, the resultant antibody recognized a 55-kDa protein in zebrafish eye extracts and in photoreceptor inner and outer segments isolated from green sunfish (Fig. 2) . The low-molecular–weight bands in the zebrafish extract are likely a result of proteolysis because they varied from preparation to preparation. Our observations confirm that the antibody cross-reacts with zebrafish proteins of the predicted molecular weight and that these proteins are concentrated in photoreceptor-enriched preparations. 
In zebrafish retinal sections immunostained for fascin 2, fluorescence was concentrated in cone inner segments in a pattern consistent with that of actin staining in this region, as visualized with an actin antibody (Figs. 3A 3B) . In the stacked confocal images, the superimposition of actin filament bundles along the sides of the ellipsoids (indicated by arrowheads in the double cone layer) stain more heavily than the individual bundles visualized en face (arrows). Though actin is similarly abundant in inner segments and the outer limiting membrane, fascin 2 staining is more specific to cone ellipsoids. A schematic of the zebrafish retina illustrates the positions of the different cone subtypes in the photoreceptor layer (Fig. 3C) . The retinal section in Figure 3is from a light-adapted zebrafish; therefore, the rods are obscured by the RPE. Preincubation of fascin 2 antibodies with a molar excess of bacterially produced zebrafish fascin 2A, zebrafish fascin 2B, or a mixture of the two eliminated staining of the cone inner segment actin filament bundles and demonstrated fascin 2 antibody specificity (data not shown). 
Isolated inner/outer segments of green sunfish photoreceptors double-labeled with the fascin 2 antibody and phalloidin also exhibited distinct colocalization of fluorescence to cone ellipsoid actin (Figs. 3D 3E 3F) . In the double-labeled preparations, cone accessory outer segments also stained intensely with the fascin 2 antibody and phalloidin (Figs. 3D 3E ; arrows in both). The accessory outer segment is a saclike structure containing the ciliary axoneme in fish cones. 18 Phalloidin labeling of cone accessory outer segments was surprising because it was not observed in our previous studies with inner/outer segment preparations, nor were actin filaments observed in cone accessory outer segments. 12 Control retinal sections stained with phalloidin alone did not stain the cone accessory outer segment, yet the fascin 2 antibody stained cone accessory outer segments in the presence or absence of phalloidin (data not shown). The strong fascin 2 labeling of accessory outer segments in the absence of actin filaments is puzzling, but we have observed accessory outer segment staining with several myosin antibodies in a variable manner (BB, unpublished observations, 2006), suggesting that these structures are prone to binding antibodies in a nonspecific manner. 
In rod inner segments, fascin 2 staining (Figs. 3G 3J)also colocalized with phalloidin labeling (Fig. 3H) . Calycal processes extending beyond the inner/outer segment junction (arrows) are clearly labeled with fascin 2 and phalloidin, though distinct inner segment actin filament bundles are difficult to discern because of the small size of the rods. Fascin 2 staining in rods does not colocalize with acetylated tubulin associated with the ciliary axoneme (Fig. 3J 3K 3L , arrowheads). The absence of fascin 2 staining in the rod ciliary axoneme, which is the homolog of the cone accessory outer segment, casts further doubt on the specificity of cone accessory outer segment staining by the fascin 2 antibody. 
The fascin 2 antibody did not appear to cross-react with X. laevis fascin 2 in spite of our inclusion of Xenopus fascin 2 fusion protein in the immunogen mix. Although the antibody did bind to bands of the appropriate molecular weight in Western blot analysis containing bacterial fusion protein and Xenopus eye extracts, no specific immunlocalization of fascin 2 was observed in retinal sections from transgenic Xenopus expressing GFP and XF2 on separate plasmids (data not shown). 
Therefore, we examined the location of Xenopus fascin 2 (XF2) in rod photoreceptors of transgenic Xenopus tadpoles using a strong rod-specific expression vector to drive expression of GFP-tagged fascins. In the rods expressing tagged fascin 2, fluorescence was clearly localized to the longitudinal ellipsoid actin bundles and calycal processes that surround the proximal outer segment (Figs. 4A 4B) . Fascin 2 staining of the more proximal part of the bundles is obscured by the abundant cytosolic fascin 2 produced by the rod opsin promoter in the transgene. Green fluorescence was also intense in the cytosol of the myoid, connecting cilium, and the axonal synapse. Similar fluorescence localization was observed in transgenic Xenopus rods expressing GFP fusions to zebrafish DrF2A and DrF2B fascins (data not shown). No morphologic defects were detectable at the light microscopic or electronic microscopic level in rods expressing the XF2 fusion protein, and no diminution of eye fluorescence was detected in older tadpoles, suggesting that expression of the fascin 2 transgene did not compromise Xenopus rod morphology or survival (data not shown). Thus, these transgenic expression studies confirm findings of our immunolocalization studies, indicating that fascin 2 colocalizes with actin filament bundles in zebrafish and Xenopus photoreceptors. As previously shown, we observed no colocalization of fluorescence with phalloidin-stained inner segment actin filament bundles in control transgenics expressing GFP alone (data not shown). 16 Rather, when GFP is expressed in transgenic Xenopus rods, fluorescence was found in rod nuclei and in all cytosolic compartments (inner segment, connecting cilium, and axonal synapse). 
In fascin proteins, serine 39 is a highly conserved residue that has been shown to regulate the binding of fascin 1 to actin. 5 In those studies, phosphorylation of serine 39 prevents fascin 1 binding to actin filaments. To investigate whether mutations in serine 39 of fascin 2 affects GFP-tagged XF2 localization in transgenic Xenopus rods, we expressed two fascin 2 serine 39 mutations in Xenopus transgenics. Substitution of neither an alanine (Figs. 4C 4D)nor an aspartic acid (to mimic a phosphorylated residue; Figs. 4E 4F ) for serine 39 of GFP/XF2 produced a consistent change in fascin 2 localization to ellipsoid actin filament bundles compared with unmutated tagged XF2. Some fluorescence colocalized with inner segment actin filament bundles could be observed in the S39A and the S39D XF2 mutants. Thus, we could not show in vivo that mutations of the serine affecting its phosphorylation state produced any obvious changes in localization. It is likely that the high level of transgene expression produced such bright cytosolic fluorescence in the inner segment that it might not have been possible to discern subtle changes in the amounts of signal colocalizing with the actin bundles expressing the three different transgenes. As with the wild-type XF2 transgene, no morphologic abnormalities were observed by light or electron microscopy in rod photoreceptors of tadpoles expressing the mutated XF2 transgenes, and no loss of eye fluorescence was detected with age (data not shown). 
Fascin 2 Actin-Binding and -Bundling Activity
To further examine the mechanism of fascin 2 association with photoreceptor actin filament bundles, we investigated fascin 2 interactions with actin filaments in vitro. Glutathione-S-transferase (GST) proteins fused to fascin 2 were purified from bacteria and incubated with polymerized nonmuscle F-actin and analyzed by sedimentation. Of the three clones used in this study, zebrafish DrF2B was the most soluble. DrF2A yielded lower levels of soluble purified protein, and XF2 was so insoluble that we were unable to obtain sufficient quantities for in vitro actin assays. Therefore, we used DrF2B to analyze fascin 2 actin-binding and actin-bundling activity in more detail. GST/DrF2B protein was incubated with polymerized actin and spun at 150,000g to separate actin filaments (pellet) from unpolymerized actin (supernatant). Significantly more GST/DrF2B was found in the pellet with actin filaments than when GST/DrF2B was incubated in the absence of F-actin (Fig. 5A) . GST alone did not pellet with actin filaments. 
We also examined the effects of phosphorylation on fascin 2 actin binding by comparing a mutant zebrafish fascin 2B with wild-type in actin-binding assays. Given that bacterially expressed proteins are not phosphorylated, wild-type GST/DrF2B represents the unphosphorylated state. To mimic phosphorylated fascin 2, we substituted an aspartic acid for serine 39 (S39D). The S39D GST/DrF2B mutant sedimented with actin filaments less effectively than wild-type GST/DrF2B; most of the S39D protein remained in the soluble fraction (Fig. 5A) . However, more GST/DrF2BS39D sedimented with actin filaments than sedimented when incubated alone, indicating that the S39D mutant does retain some actin-binding activity at markedly reduced levels. These results are consistent with those found with fascin 1, suggesting that fascin 2 serine 39 phosphorylation reduces, but does not abolish, its affinity for actin filaments. 
Because zebrafish have two fascin 2 genes, we were interested in comparing the actin-binding activities of the two isoforms. Slightly more GST/DrF2B sedimented with actin filaments than GST/DrF2A (Fig. 5B) . However, the presence of low-molecular–weight bands in the GST/DrF2A samples suggested that protein degradation might have contributed to the reduced actin-binding. Furthermore, GST/DrF2A protein was consistently less soluble than GST/DrF2B. Because yields for the GST/DrF2A fusion protein were considerably less than for GST/DrF2B, concentration of the protein was required before incubation with actin. This additional handling likely increased protein degradation and contributed to the reduced actin-binding activity. 
To examine the binding affinities of the two zebrafish wild-type fascin 2 and the S39D mutant fascin 2B, molar ratios of bound fusion protein to actin filaments were determined by incubating actin filaments with increasing concentrations of GST/fascin 2 fusion proteins (Fig. 6) . Mutation of serine 39 to aspartic acid to mimic a phosphorylated serine in fascin 2B dramatically reduced, but did not abolish, actin filament binding. The binding affinity of GST/DrF2A for actin filaments was lower than that for GST/DrF2B but higher than that for GST/DrF2BS39D. 
The ability of wild-type and mutant zebrafish fascin 2 fusion proteins to bundle actin filaments was also examined by sedimentation assays. A low-speed spin of fusion protein and polymerized actin mixtures sediments actin only when actin filaments have been bundled. In samples without fusion protein, very low levels of actin sedimented into the pellet (Fig. 7A) . When GST/DrF2B was incubated with polymerized actin, most of the actin sedimented into the low-speed pellet because of the formation of actin filament bundles. In addition, more GST/DrF2B sedimented to the pellet when incubated with actin than when incubated alone. The addition of GST alone to actin filaments did not shift the fusion protein or actin to the pellet. The actin filament bundles formed by GST/DrF2B were visualized using fluorescently tagged phalloidin (Fig. 8) . A meshwork of brightly labeled actin bundles was observed in the presence of GST/DrF2B (Fig. 8B)and differed from the labeling of actin filaments alone (Fig. 8A)or of F-actin when incubated with GST alone (Fig. 8C) . Some smaller protein aggregates formed in the presence of GST but were less ordered than the bundles formed in the presence of fascin 2. The GST/DrF2BS39D mutant was also tested for its ability to bundle actin filaments by sedimentation assays, and no change occurred in the amount of actin pelleted when GST/DrF2BS39D was added, indicating that the S39D mutant is unable to bundle actin (Fig. 7B) . Thus, it appears that although the S39D mutant fascin 2 retains a modest level of actin filament-binding activity, it is not able to bundle actin. GST/DrF2A was also analyzed for actin filament-bundling activity (Fig. 7C) . Incubation of GST/DrF2A with actin filaments produced no change in the amount of actin or fusion protein sedimented to the pellet after centrifugation, suggesting that this zebrafish isoform lacks actin filament-bundling activity. 
The actin-bundling activity of both zebrafish fascin 2 proteins was compared in detail by examining the amount of actin filaments bundled as a function of added fusion protein concentration (Fig. 9) . Although DrF2B exhibited robust actin-bundling activity, DrF2A was unable to bundle actin filaments effectively at any of the concentrations examined. These observations indicate that the GST/DrF2A fusion protein can bind actin filaments with reduced affinity compared with GST/DrF2B, but it is unable to organize the filaments into bundles. 
Discussion
Our objective in this study was to elucidate the function of fascin 2 in photoreceptors by obtaining a more detailed understanding of fascin 2 subcellular localization and examining its ability to bind and bundle actin filaments. The fascin 2 cDNA cloned from X. laevis and zebrafish retinal libraries for this study exhibit extensive sequence conservation with previously characterized fascin 2 genes and other fascin family members. The presence of two fascin 2 genes in zebrafish most likely reflects the previously described genome duplication that occurred in ray-finned fish during evolution. 19 There are two fascin 1 and fascin 2 genes in zebrafish and Fugu; in most other vertebrates, fascins are represented by only a single gene. It is not yet possible to ascertain whether the two fascin 2 proteins have evolved distinct functions in fish photoreceptors. The marked difference in solubility of the two zebrafish fascin 2 proteins, when expressed as fusion proteins in bacteria, suggests that these proteins do have different structural properties. Because the antibody we generated for this study recognizes DrF2A and DrF2B, we cannot ascertain whether both proteins are expressed in photoreceptors or if they have different localizations. Our studies show that DrF2B has greater affinity for actin filaments and much greater actin filament–bundling activity than DrF2A. Indeed, DrF2A exhibits little, if any, bundling activity. This discrepancy between the two proteins may reflect the quality of the purified protein rather than differences in their ability to interact with actin filaments. 
With the use of immunohistochemistry in zebrafish retina and fluorescently tagged fascin 2 in transgenic Xenopus tadpole retinas, we showed that fascin 2 colocalizes with photoreceptor inner segment actin filament bundles and calycal processes. Fascin 2 immunolocalization in preparations of isolated cone inner/outer segments from sunfish retina provides sufficient resolution to make it clear that the fascin 2 antibody stains the actin filament bundles in the ellipsoid and calycal processes uniformly along the proximal–distal axis. Fascin 2 immunostaining almost exactly colocalizes with phalloidin staining of actin filament bundles. This colocalization of fascin 2 with inner segment actin bundles suggests that fascin 2 plays a role in bundling the inner segment actin filaments, analogous to the structural role of fascin 1 in bundling the actin filaments of filopodia. Fascin 1 is required for the formation of filopodia-like actin bundles in an in vitro system and has been hypothesized to cross-link and stabilize filopodial actin filament bundles, thereby generating sufficient structural stiffness to support filopodial protrusion. 20 A similar structural role for fascin 2 in the generation and maintenance of calycal processes would be consistent with our observations. 
Saishin et al. 8 reported that fascin 2 was localized to inner and outer segments by immunohistochemistry of bovine retina and suggested that fascin 2 mutation in humans may disrupt outer segment morphogenesis by interfering with actin filament function in the distal connecting cilium. Mice homozygous for mutated fascin 2 also had misaligned outer segment disks near the distal connecting cilium and bent outer segments. 11 However, in our studies, localization to the distal connecting cilium or the outer segment was not observed. We cannot rule out the possibility that fascin 2 is present at this location at concentrations below our level of detection. However, because the actin filaments in the connecting cilium are not bundled and because fascin 2 appears to function like fascin 1 as an actin filament-bundling protein, we think it unlikely that the connecting cilium is the primary locus of fascin 2 function in photoreceptors. 
Our findings suggest that fascin 2 actin filament–binding and –bundling activity are regulated by the phosphorylation of serine 39 of fascin 2, as has been reported for fascin 1. 5 21 In sedimentation assays in vitro, a mutation that mimics serine 39 phosphorylation resulted in a severe reduction in actin-binding compared with unphosphorylated fascin 2, as has been observed for fascin 1. The affinity of the S39D mutant for actin filaments was reduced, but not eliminated, in these assays. When we examined the same mutation in vivo with GFP-tagged Xenopus fascin 2 transgenes, mutant and wild-type fascin 2 proteins localized to actin filament bundles. It is likely that the transgenic system fails to effectively detect relative differences in actin filament binding affinity because the rod opsin promoter expresses such high levels of fascin 2. Fascin 2 fusion protein fluorescence in the soluble cytosolic compartment of the proximal inner segment is often so intense that it obscures the fluorescence associated with actin bundles in this region. Such elevated fascin 2 levels are likely to saturate fascin 2 binding sites on actin filaments, with excess fascin 2 remaining in the rod cytosol. 
It was surprising that no change in photoreceptor morphology was detected (i.e., no obvious changes in inner segment actin bundle size or arrangement and no change in the length or width of calycal processes) in fascin 2 and fascin 2 mutant serine 39 transgenic Xenopus rod photoreceptors because a strong rod opsin promoter was used and might have been expected to overexpress fascin 2, thereby disrupting the actin cytoskeleton. There are several possible explanations for these observations. Photoreceptors may be able to regulate actin bundle assembly in the inner segment and calycal processes despite the presence of excess fascin 2. Alternatively, the fusion of GFP to fascin 2 may allow actin filament binding but may interfere with actin filament-bundling activity; this seems likely since cross-linking filaments into bundles is spatially constrained. Another possibility is that endogenous fascin 2 is already expressed at high enough concentrations in rods that its bundling activity is saturated. Turnover of inner segment actin bundles may also be slow enough that bundles formed before the onset of opsin expression remain stable. 
The haploinsufficiency of fascin 2 in patients with autosomal dominant retinitis pigmentosa and macular degeneration and containing a single copy of the wild-type gene suggests that reduced amounts of fascin 2 are produced in photoreceptors of these patients and that reduced levels fail to maintain photoreceptors. Given that photoreceptors develop and then degenerate in human fascin 2 heterozygotes, expression from a single copy of the fascin 2 gene appears to be sufficient to support photoreceptor morphogenesis. In mice homozygous for a fascin 2 mutation or a disrupted fascin 2 gene, photoreceptors also develop normally and degenerate later, demonstrating that fascin 2 is not in fact required for photoreceptor morphogenesis. However, the very slender mouse photoreceptors differ from those of most vertebrates in that they lack calycal processes and lack actin bundles in their inner segments. It is unknown what effect the absence of fascin 2 would have in more typical vertebrate photoreceptors with calycal processes and inner segment actin filament bundles such as those in humans. 
Several observations suggest that fascin 2 is not likely to be the only actin filament cross-linking protein associated with inner segment actin filaments and calycal processes in vertebrate photoreceptors. For example, a second cross-linking protein might normally work in concert with fascin 2 during normal photoreceptor development but nonetheless have sufficient actin-bundling activity of its own to support photoreceptor morphogenesis when fascin 2 protein levels are reduced by haploinsufficiency in humans. A candidate for such a second actin filament cross-linker is fimbrin (also called plastin), which is also localized to actin filament bundles in photoreceptor inner segments and calycal processes. 22 There are several examples in the literature in which the assembly of parallel, polarized actin filament bundles is mediated by the collective interaction of multiple actin-bundling proteins with overlapping functions. Examples include espin and T-plastin in hair cell stereocilia, 23 24 fascin (singed gene product) and forked (espin homolog) during Drosophila bristle development, 25 and fascin and quail (villin-like protein) in Drosophila nurse cells. 26 27 In Drosophila bristles, forked acts before fascin to organize actin filaments into small bundles, and then fascin further cross-links these small bundles into large ones, thereby imparting stiffness to the bristles. 28 Thus, the timing of action and the unique bundling properties of multiple actin–cross-linking proteins can contribute to the ultimate actin filament organization and overall structure of the resultant bundles. Orchestrating the participation of multiple cross-linkers is likely to play a critical role in cell morphogenesis. 
In summary, we have shown that fascin 2 binds and bundles actin filaments and that it is localized to the actin filament bundles of photoreceptor inner segments and calycal processes. These observations suggest that reducing fascin 2 levels in photoreceptors might be expected to compromise the structural integrity of the inner segment actin cytoskeleton of photoreceptors. Reduced stability of inner segment actin filament bundles in photoreceptors of patients with fascin 2 mutations may lead over time to compromised intracellular transport or compromised critical structural relationships. It will be interesting to analyze the structure of the inner segment bundles in photoreceptors that have reduced levels of fascin 2. 
 
Figure 1.
 
ClustalW alignment of fascin 2 amino acid sequences. Fascin 2 protein sequences from X. laevis (XlFII), D. rerio (DrFIIA and DrFIIB), human (HsFII), cow (BtFII), mouse (MmFII), rat (RnFII), Xenopus tropicalis (XtFII), and Fugu rubripes (FrFII) were compared. Residues outlined in black are identical between species, and residues outlined in gray are conserved. For the consensus, an asterisk indicates an amino acid found in all fascin 2 sequences listed, and a period denotes a conserved residue. The MARCKS homology domain, which likely contains the N-terminal actin-binding site, is indicated by brackets above the sequence and contains the highly conserved serine 39 residue found in almost all fascins (marked by an asterisk).
Figure 1.
 
ClustalW alignment of fascin 2 amino acid sequences. Fascin 2 protein sequences from X. laevis (XlFII), D. rerio (DrFIIA and DrFIIB), human (HsFII), cow (BtFII), mouse (MmFII), rat (RnFII), Xenopus tropicalis (XtFII), and Fugu rubripes (FrFII) were compared. Residues outlined in black are identical between species, and residues outlined in gray are conserved. For the consensus, an asterisk indicates an amino acid found in all fascin 2 sequences listed, and a period denotes a conserved residue. The MARCKS homology domain, which likely contains the N-terminal actin-binding site, is indicated by brackets above the sequence and contains the highly conserved serine 39 residue found in almost all fascins (marked by an asterisk).
Figure 2.
 
A fascin 2 antibody recognizes a 55-kDa band in extracts from zebrafish eyes and green sunfish photoreceptor inner segment/outer segment preparations. An affinity-purified antibody produced against Xenopus and zebrafish fascin 2 fusion proteins detects a 55-kDa protein (arrow) in extracts from whole zebrafish eyes and green sunfish preparations enriched in photoreceptor inner and outer segment fragments. The low-molecular–weight bands seen in the zebrafish sample most likely represent degradation products of zebrafish fascin 2.
Figure 2.
 
A fascin 2 antibody recognizes a 55-kDa band in extracts from zebrafish eyes and green sunfish photoreceptor inner segment/outer segment preparations. An affinity-purified antibody produced against Xenopus and zebrafish fascin 2 fusion proteins detects a 55-kDa protein (arrow) in extracts from whole zebrafish eyes and green sunfish preparations enriched in photoreceptor inner and outer segment fragments. The low-molecular–weight bands seen in the zebrafish sample most likely represent degradation products of zebrafish fascin 2.
Figure 3.
 
Fascin 2 colocalizes with photoreceptor inner segment actin filament bundles in fish retina. Fascin 2 staining (A) in zebrafish cone inner segments colocalizes with actin antibody staining (B) of inner segment actin filament bundles that run the length of the cone ellipsoids (arrows). (C) Schematic of the zebrafish outer nuclear layer illustrates the positions of the different cone subtypes. RPE, retinal pigmented epithelium; R, rod layer; DC, double-cone layer; LSC, long single-cone layer; SSC, short single-cone layer; OLM, outer limiting membrane; RN, rod nuclear layer. Isolated photoreceptor inner/outer segments from green sunfish retinas were stained with α-fascin 2 (D, G, J), Alexa Fluor 488 phalloidin (E, H), and α-acetylated tubulin (K) and visualized in DIC (F, I, M). A double-cone inner segment fragment contains accessory outer segments (arrows) and truncated outer segments (OS) (DF). Fascin 2 and phalloidin exhibit almost identical staining patterns of the cone inner segment actin filament bundles and accessory outer segment. Fascin 2 antibodies and phalloidin stain the rod inner segment but not the outer segment (GI). Staining is reduced in the mitochondrial-rich ellipsoids located just below the inner/outer segment junction (arrows). Fascin 2 and phalloidin labeling extends beyond the inner/outer segment junction into the calycal processes. Fascin 2 staining of the rod inner segment and calycal processes (arrows) does not colocalize with the ciliary axoneme (arrowhead) labeled with an acetylated tubulin antibody (JM). Scale bar: (A, B) 20 μm; (DM) 5 μm.
Figure 3.
 
Fascin 2 colocalizes with photoreceptor inner segment actin filament bundles in fish retina. Fascin 2 staining (A) in zebrafish cone inner segments colocalizes with actin antibody staining (B) of inner segment actin filament bundles that run the length of the cone ellipsoids (arrows). (C) Schematic of the zebrafish outer nuclear layer illustrates the positions of the different cone subtypes. RPE, retinal pigmented epithelium; R, rod layer; DC, double-cone layer; LSC, long single-cone layer; SSC, short single-cone layer; OLM, outer limiting membrane; RN, rod nuclear layer. Isolated photoreceptor inner/outer segments from green sunfish retinas were stained with α-fascin 2 (D, G, J), Alexa Fluor 488 phalloidin (E, H), and α-acetylated tubulin (K) and visualized in DIC (F, I, M). A double-cone inner segment fragment contains accessory outer segments (arrows) and truncated outer segments (OS) (DF). Fascin 2 and phalloidin exhibit almost identical staining patterns of the cone inner segment actin filament bundles and accessory outer segment. Fascin 2 antibodies and phalloidin stain the rod inner segment but not the outer segment (GI). Staining is reduced in the mitochondrial-rich ellipsoids located just below the inner/outer segment junction (arrows). Fascin 2 and phalloidin labeling extends beyond the inner/outer segment junction into the calycal processes. Fascin 2 staining of the rod inner segment and calycal processes (arrows) does not colocalize with the ciliary axoneme (arrowhead) labeled with an acetylated tubulin antibody (JM). Scale bar: (A, B) 20 μm; (DM) 5 μm.
Figure 4.
 
Localization of wild-type and mutant fascin 2 proteins tagged with green fluorescent protein (GFP) in transgenic Xenopus tadpole rods. (A) In transgenic Xenopus tadpole retina, GFP-tagged XF2 is abundant in rod cytosolic compartments and colocalizes with actin filament bundles in the ellipsoid and calycal processes (arrows). Fluorescence is also bright in the axonal synapse at the outer plexiform layer (OPL). (B) The inner segment actin filament bundles (arrows) that extend into the calycal processes near the proximal outer segment (OS) are also labeled with Texas-Red phalloidin. A similar pattern of transgenic protein localization was found in rods containing serine 39 mutations to an alanine (C) or an aspartic acid (E). Fluorescence was predominantly in the rod cytosolic compartments, including the connecting cilium that can be seen in rods of the GFP/XF2S39D retina (E, arrowheads). However, some fluorescence colocalizing with actin filament bundles visualized with phalloidin (D, F) was also observed. OLM, outer limiting membrane. Scale bar: 10 μm.
Figure 4.
 
Localization of wild-type and mutant fascin 2 proteins tagged with green fluorescent protein (GFP) in transgenic Xenopus tadpole rods. (A) In transgenic Xenopus tadpole retina, GFP-tagged XF2 is abundant in rod cytosolic compartments and colocalizes with actin filament bundles in the ellipsoid and calycal processes (arrows). Fluorescence is also bright in the axonal synapse at the outer plexiform layer (OPL). (B) The inner segment actin filament bundles (arrows) that extend into the calycal processes near the proximal outer segment (OS) are also labeled with Texas-Red phalloidin. A similar pattern of transgenic protein localization was found in rods containing serine 39 mutations to an alanine (C) or an aspartic acid (E). Fluorescence was predominantly in the rod cytosolic compartments, including the connecting cilium that can be seen in rods of the GFP/XF2S39D retina (E, arrowheads). However, some fluorescence colocalizing with actin filament bundles visualized with phalloidin (D, F) was also observed. OLM, outer limiting membrane. Scale bar: 10 μm.
Figure 5.
 
Zebrafish fascin 2A and 2B bind actin filaments. (A) Purified GST/DrF2B fusion proteins sediment with actin filaments in the high-speed pellet (hsp). Mutation of serine 39 to an aspartic acid reduces the amount of fusion protein in the pellet. In the absence of actin, little of the fusion proteins is found in the high-speed pellet. GST alone does not bind to actin and is not found in the high-speed pellet in the absence or presence of actin. (B) A greater amount of the zebrafish fascin 2B fusion protein sediments with F-actin in the high-speed pellet compared to zebrafish fascin 2A. Fascin 2A fusion protein exhibits additional low-molecular–weight bands in the high-speed supernatant (hss), possibly indicating the presence of proteolytic fragments that do not bind actin filaments efficiently.
Figure 5.
 
Zebrafish fascin 2A and 2B bind actin filaments. (A) Purified GST/DrF2B fusion proteins sediment with actin filaments in the high-speed pellet (hsp). Mutation of serine 39 to an aspartic acid reduces the amount of fusion protein in the pellet. In the absence of actin, little of the fusion proteins is found in the high-speed pellet. GST alone does not bind to actin and is not found in the high-speed pellet in the absence or presence of actin. (B) A greater amount of the zebrafish fascin 2B fusion protein sediments with F-actin in the high-speed pellet compared to zebrafish fascin 2A. Fascin 2A fusion protein exhibits additional low-molecular–weight bands in the high-speed supernatant (hss), possibly indicating the presence of proteolytic fragments that do not bind actin filaments efficiently.
Figure 6.
 
Actin filament-binding activities of zebrafish fascin 2A and 2B and a S39D mutant of fascin 2B. Polymerized nonmuscle actin (7.1 μM) was combined with increasing concentrations of GST fusion proteins of zebrafish DrF2A, DrF2B, and the DrF2B serine 39 mutation to an aspartic acid. The amount of fusion protein associated with actin filaments relative to the amount of polymerized actin was determined by analyzing samples fractionated by centrifugation and visualized by SDS-PAGE. DrF2B had the greatest affinity for actin filaments with a significant loss of actin-binding activity when the conserved serine 39 was mutated to an aspartic acid to mimic a phosphorylated residue. The DrF2BS39D mutant does retain some residual low level of actin filament binding. DrF2A has a lower affinity for actin filament-binding compared with DrF2B but was also more insoluble during purification. However, DrF2A still has a greater affinity for actin filaments than the DrF2BS39D mutant. Error bars represent SD from at least four different samples.
Figure 6.
 
Actin filament-binding activities of zebrafish fascin 2A and 2B and a S39D mutant of fascin 2B. Polymerized nonmuscle actin (7.1 μM) was combined with increasing concentrations of GST fusion proteins of zebrafish DrF2A, DrF2B, and the DrF2B serine 39 mutation to an aspartic acid. The amount of fusion protein associated with actin filaments relative to the amount of polymerized actin was determined by analyzing samples fractionated by centrifugation and visualized by SDS-PAGE. DrF2B had the greatest affinity for actin filaments with a significant loss of actin-binding activity when the conserved serine 39 was mutated to an aspartic acid to mimic a phosphorylated residue. The DrF2BS39D mutant does retain some residual low level of actin filament binding. DrF2A has a lower affinity for actin filament-binding compared with DrF2B but was also more insoluble during purification. However, DrF2A still has a greater affinity for actin filaments than the DrF2BS39D mutant. Error bars represent SD from at least four different samples.
Figure 7.
 
A zebrafish GST–fascin 2B fusion protein bundles actin filaments. (A) When spun at low speed, filamentous actin remains in the supernatant (lss). If DrF2B fusion protein is added, actin filament bundles sediment into the low-speed pellet (lsp), and more zebrafish fascin 2B fusion protein sediments into the pellet than when the fusion protein is sedimented alone. GST alone does not bundle actin; the amount of actin filaments in the low-speed pellet remains unchanged in the presence or absence of GST. (B) If serine 39 is mutated to an aspartic acid residue in zebrafish fascin 2B fusion protein, sedimentation of actin filaments is lost. (C) Zebrafish fascin 2A fusion protein has dramatically less actin-bundling activity than zebrafish fascin 2B fusion protein. No change in actin filament sedimentation is observed when zebrafish fascin 2A fusion protein is added.
Figure 7.
 
A zebrafish GST–fascin 2B fusion protein bundles actin filaments. (A) When spun at low speed, filamentous actin remains in the supernatant (lss). If DrF2B fusion protein is added, actin filament bundles sediment into the low-speed pellet (lsp), and more zebrafish fascin 2B fusion protein sediments into the pellet than when the fusion protein is sedimented alone. GST alone does not bundle actin; the amount of actin filaments in the low-speed pellet remains unchanged in the presence or absence of GST. (B) If serine 39 is mutated to an aspartic acid residue in zebrafish fascin 2B fusion protein, sedimentation of actin filaments is lost. (C) Zebrafish fascin 2A fusion protein has dramatically less actin-bundling activity than zebrafish fascin 2B fusion protein. No change in actin filament sedimentation is observed when zebrafish fascin 2A fusion protein is added.
Figure 8.
 
Visualization of actin filament bundles cross-linked by zebrafish fascin 2B using fluorescence-tagged phalloidin. (A) Filamentous actin labeled with Alexa Fluor 488 phalloidin. (B) Adding 4 μM of a GST fusion to zebrafish fascin 2B resulted in a meshwork of thick, bundled actin filaments. (C) GST (4 μM) did not bundle actin filaments, though there might have been some protein aggregation. Scale bar: 5 μm.
Figure 8.
 
Visualization of actin filament bundles cross-linked by zebrafish fascin 2B using fluorescence-tagged phalloidin. (A) Filamentous actin labeled with Alexa Fluor 488 phalloidin. (B) Adding 4 μM of a GST fusion to zebrafish fascin 2B resulted in a meshwork of thick, bundled actin filaments. (C) GST (4 μM) did not bundle actin filaments, though there might have been some protein aggregation. Scale bar: 5 μm.
Figure 9.
 
Analysis of zebrafish fascin 2 actin filament-bundling activity. Increasing concentrations of GST/DrF2A and GST/DrF2B were incubated with 7.1 μM polymerized actin, and samples were analyzed for actin-bundling activity by low-speed centrifugation and SDS-PAGE. With increasing concentrations of GST/DrF2B, greater percentages of actin-filaments were observed in bundles found in the low-speed pellet. DrF2A showed bundling activity only at the highest concentrations tested. Error bars represent the SD from four or more samples.
Figure 9.
 
Analysis of zebrafish fascin 2 actin filament-bundling activity. Increasing concentrations of GST/DrF2A and GST/DrF2B were incubated with 7.1 μM polymerized actin, and samples were analyzed for actin-bundling activity by low-speed centrifugation and SDS-PAGE. With increasing concentrations of GST/DrF2B, greater percentages of actin-filaments were observed in bundles found in the low-speed pellet. DrF2A showed bundling activity only at the highest concentrations tested. Error bars represent the SD from four or more samples.
The authors thank Jennifer Fogarty, Ed Parker, and Mike Wu for their technical assistance, Andréa Dosé and Amy Corsa for comments on the manuscript, and other members of the Burnside laboratory for their useful input. 
KureishyN, SapountziV, PragS, AnilkumarN, AdamsJC. Fascins, and their roles in cell structure and function. Bioessays. 2002;24:350–361. [CrossRef] [PubMed]
TubbB, MulhollandDJ, VoglW, et al. Testis fascin (FSCN3): a novel paralog of the actin-bundling protein fascin expressed specifically in the elongate spermatid head. Exp Cell Res. 2002;275:92–109. [CrossRef] [PubMed]
Yamashiro-MatsumuraS, MatsumuraF. Intracellular localization of the 55-kD actin-bundling protein in cultured cells: spatial relationships with actin, alpha-actinin, tropomyosin, and fimbrin. J Cell Biol. 1986;103:631–640. [CrossRef] [PubMed]
Yamashiro-MatsumuraS, MatsumuraF. Purification and characterization of an F-actin-bundling 55-kilodalton protein from HeLa cells. J Biol Chem. 1985;260:5087–5097. [PubMed]
OnoS, YamakitaY, YamashiroS, et al. Identification of an actin binding region and a protein kinase C phosphorylation site on human fascin. J Biol Chem. 1997;272:2527–2533. [CrossRef] [PubMed]
YamakitaY, OnoS, MatsumuraF, YamashiroS. Phosphorylation of human fascin inhibits its actin binding and bundling activities. J Biol Chem. 1996;271:12632–12638. [CrossRef] [PubMed]
AdamsJC, ClellandJD, CollettGD, MatsumuraF, YamashiroS, ZhangL. Cell-matrix adhesions differentially regulate fascin phosphorylation. Mol Biol Cell. 1999;10:4177–4190. [CrossRef] [PubMed]
SaishinY, IshikawaR, UgawaS, et al. Retinal fascin: functional nature, subcellular distribution, and chromosomal localization. Invest Ophthalmol Vis Sci. 2000;41:2087–2095. [PubMed]
WadaY, AbeT, ItabashiT, SatoH, KawamuraM, TamaiM. Autosomal dominant macular degeneration associated with 208delG mutation in the FSCN2 gene. Arch Ophthalmol. 2003;121:1613–1620. [CrossRef] [PubMed]
WadaY, AbeT, TakeshitaT, SatoH, YanashimaK, TamaiM. Mutation of human retinal fascin gene (FSCN2) causes autosomal dominant retinitis pigmentosa. Invest Ophthalmol Vis Sci. 2001;42:2395–2400. [PubMed]
YokokuraS, WadaY, NakaiS, et al. Targeted disruption of FSCN2 gene induces retinopathy in mice. Invest Ophthalmol Vis Sci. 2005;46:2905–2915. [CrossRef] [PubMed]
NagleBW, OkamotoC, TaggartB, BurnsideB. The teleost cone cytoskeleton: localization of actin, microtubules, and intermediate filaments. Invest Ophthalmol Vis Sci. 1986;27:689–701. [PubMed]
ChaitinMH, SchneiderBG, HallMO, PapermasterDS. Actin in the photoreceptor connecting cilium: immunocytochemical localization to the site of outer segment disk formation. J Cell Biol. 1984;99:239–247. [CrossRef] [PubMed]
WilliamsDS, LinbergKA, VaughnDK, FarissRN, FisherSK. Disruption of microfilament organization and deregulation of disk membrane morphogenesis by cytochalisin D in rod and cone photoreceptors. J Comp Neurol. 1988;272:161–176. [CrossRef] [PubMed]
Lin-JonesJ, ParkerE, WuM, KnoxBE, BurnsideB. Disruption of kinesin II function using a dominant negative-acting transgene in Xenopus laevis rods results in photoreceptor degeneration. Invest Ophthalmol Vis Sci. 2003;44:3614–3621. [CrossRef] [PubMed]
Lin-JonesJ, ParkerE, WuM, DoséA, BurnsideB. Myosin 3A transgene expression produces abnormal actin filament bundles in transgenic Xenopus laevis rod photoreceptors. J Cell Sci. 2004;117:5825–5834. [CrossRef] [PubMed]
DoséAC, HillmanDW, WongC, SohlbergL, Lin-JonesJ, BurnsideB. Myo3A, one of two class III myosin genes expressed in vertebrate retina, is localized to the calycal processes of rod and cone photoreceptors and is expressed in the sacculus. Mol Biol Cell. 2003;14:1058–1073. [CrossRef] [PubMed]
YacobA, WiseC, KunzYW. The accessory outer segment of rods and cones in the retina of the guppy, Poecilia reticulata P. (Teleostei): an electron microscopical study. Cell Tissue Res. 1977;177:181–193. [PubMed]
MeyerA, Van de PeerY. From 2R to 3R: evidence for a fish-specific genome duplication (FSGD). Bioessays. 2005;27:937–945. [CrossRef] [PubMed]
VignjevicD, YararD, WelchMD, PeloquinJ, SvitkinaT, BorisyGG. Formation of filopodia-like bundles in vitro from a dendritic network. J Cell Biol. 2003;160:951–962. [CrossRef] [PubMed]
AnilkumarN, ParsonsM, MonkR, NgT, AdamsJC. Interaction of fascin and protein kinase Cα: a novel intersection in cell adhesion and motility. EMBO J. 2003;22:5390–5402. [CrossRef] [PubMed]
HöferD, DrenckhahnD. Molecular heterogeneity of the actin filament cytoskeleton associated with microvilli of photoreceptors, Müller’s glial cells and pigment epithelial cells of the retina. Histochemistry. 1993;99:29–35. [CrossRef] [PubMed]
DaudetN, LebartMC. Transient expression of the t-isoform of plastins/fimbrin in the stereocilia of developing auditory hair cells. Cell Motil Cytoskeleton. 2002;53:326–336. [CrossRef] [PubMed]
ZhengL, SekerkovaG, VranichK, TilneyLG, MugnainiE, BartlesJR. The deaf jerker mouse has a mutation in the gene encoding the espin actin-bundling proteins of hair cell stereocilia and lacks espins. Cell. 2000;102:377–385. [CrossRef] [PubMed]
TilneyLG, TilneyMS, GuildGM. F actin bundles in Drosophila bristles, I: two filament cross-links are involved in bundling. J Cell Biol. 1995;130:629–638. [CrossRef] [PubMed]
CantK, KnowlesBA, MoosekerMS, CooleyL. Drosophila singed, a fascin homolog, is required for actin bundle formation during oogenesis and bristle extension. J Cell Biol. 1994;125:369–380. [CrossRef] [PubMed]
Mahajan-MiklosS, CooleyL. The villin-like protein encoded by the Drosophila quail gene is required for actin bundle assembly during oogenesis. Cell. 1994;78:291–301. [CrossRef] [PubMed]
TilneyLG, ConnellyPS, VranichKA, ShawMK, GuildGM. Why are two different cross-linkers necessary for actin bundle formation in vivo and what does each cross-link contribute?. J Cell Biol. 1998;143:121–133. [CrossRef] [PubMed]
Figure 1.
 
ClustalW alignment of fascin 2 amino acid sequences. Fascin 2 protein sequences from X. laevis (XlFII), D. rerio (DrFIIA and DrFIIB), human (HsFII), cow (BtFII), mouse (MmFII), rat (RnFII), Xenopus tropicalis (XtFII), and Fugu rubripes (FrFII) were compared. Residues outlined in black are identical between species, and residues outlined in gray are conserved. For the consensus, an asterisk indicates an amino acid found in all fascin 2 sequences listed, and a period denotes a conserved residue. The MARCKS homology domain, which likely contains the N-terminal actin-binding site, is indicated by brackets above the sequence and contains the highly conserved serine 39 residue found in almost all fascins (marked by an asterisk).
Figure 1.
 
ClustalW alignment of fascin 2 amino acid sequences. Fascin 2 protein sequences from X. laevis (XlFII), D. rerio (DrFIIA and DrFIIB), human (HsFII), cow (BtFII), mouse (MmFII), rat (RnFII), Xenopus tropicalis (XtFII), and Fugu rubripes (FrFII) were compared. Residues outlined in black are identical between species, and residues outlined in gray are conserved. For the consensus, an asterisk indicates an amino acid found in all fascin 2 sequences listed, and a period denotes a conserved residue. The MARCKS homology domain, which likely contains the N-terminal actin-binding site, is indicated by brackets above the sequence and contains the highly conserved serine 39 residue found in almost all fascins (marked by an asterisk).
Figure 2.
 
A fascin 2 antibody recognizes a 55-kDa band in extracts from zebrafish eyes and green sunfish photoreceptor inner segment/outer segment preparations. An affinity-purified antibody produced against Xenopus and zebrafish fascin 2 fusion proteins detects a 55-kDa protein (arrow) in extracts from whole zebrafish eyes and green sunfish preparations enriched in photoreceptor inner and outer segment fragments. The low-molecular–weight bands seen in the zebrafish sample most likely represent degradation products of zebrafish fascin 2.
Figure 2.
 
A fascin 2 antibody recognizes a 55-kDa band in extracts from zebrafish eyes and green sunfish photoreceptor inner segment/outer segment preparations. An affinity-purified antibody produced against Xenopus and zebrafish fascin 2 fusion proteins detects a 55-kDa protein (arrow) in extracts from whole zebrafish eyes and green sunfish preparations enriched in photoreceptor inner and outer segment fragments. The low-molecular–weight bands seen in the zebrafish sample most likely represent degradation products of zebrafish fascin 2.
Figure 3.
 
Fascin 2 colocalizes with photoreceptor inner segment actin filament bundles in fish retina. Fascin 2 staining (A) in zebrafish cone inner segments colocalizes with actin antibody staining (B) of inner segment actin filament bundles that run the length of the cone ellipsoids (arrows). (C) Schematic of the zebrafish outer nuclear layer illustrates the positions of the different cone subtypes. RPE, retinal pigmented epithelium; R, rod layer; DC, double-cone layer; LSC, long single-cone layer; SSC, short single-cone layer; OLM, outer limiting membrane; RN, rod nuclear layer. Isolated photoreceptor inner/outer segments from green sunfish retinas were stained with α-fascin 2 (D, G, J), Alexa Fluor 488 phalloidin (E, H), and α-acetylated tubulin (K) and visualized in DIC (F, I, M). A double-cone inner segment fragment contains accessory outer segments (arrows) and truncated outer segments (OS) (DF). Fascin 2 and phalloidin exhibit almost identical staining patterns of the cone inner segment actin filament bundles and accessory outer segment. Fascin 2 antibodies and phalloidin stain the rod inner segment but not the outer segment (GI). Staining is reduced in the mitochondrial-rich ellipsoids located just below the inner/outer segment junction (arrows). Fascin 2 and phalloidin labeling extends beyond the inner/outer segment junction into the calycal processes. Fascin 2 staining of the rod inner segment and calycal processes (arrows) does not colocalize with the ciliary axoneme (arrowhead) labeled with an acetylated tubulin antibody (JM). Scale bar: (A, B) 20 μm; (DM) 5 μm.
Figure 3.
 
Fascin 2 colocalizes with photoreceptor inner segment actin filament bundles in fish retina. Fascin 2 staining (A) in zebrafish cone inner segments colocalizes with actin antibody staining (B) of inner segment actin filament bundles that run the length of the cone ellipsoids (arrows). (C) Schematic of the zebrafish outer nuclear layer illustrates the positions of the different cone subtypes. RPE, retinal pigmented epithelium; R, rod layer; DC, double-cone layer; LSC, long single-cone layer; SSC, short single-cone layer; OLM, outer limiting membrane; RN, rod nuclear layer. Isolated photoreceptor inner/outer segments from green sunfish retinas were stained with α-fascin 2 (D, G, J), Alexa Fluor 488 phalloidin (E, H), and α-acetylated tubulin (K) and visualized in DIC (F, I, M). A double-cone inner segment fragment contains accessory outer segments (arrows) and truncated outer segments (OS) (DF). Fascin 2 and phalloidin exhibit almost identical staining patterns of the cone inner segment actin filament bundles and accessory outer segment. Fascin 2 antibodies and phalloidin stain the rod inner segment but not the outer segment (GI). Staining is reduced in the mitochondrial-rich ellipsoids located just below the inner/outer segment junction (arrows). Fascin 2 and phalloidin labeling extends beyond the inner/outer segment junction into the calycal processes. Fascin 2 staining of the rod inner segment and calycal processes (arrows) does not colocalize with the ciliary axoneme (arrowhead) labeled with an acetylated tubulin antibody (JM). Scale bar: (A, B) 20 μm; (DM) 5 μm.
Figure 4.
 
Localization of wild-type and mutant fascin 2 proteins tagged with green fluorescent protein (GFP) in transgenic Xenopus tadpole rods. (A) In transgenic Xenopus tadpole retina, GFP-tagged XF2 is abundant in rod cytosolic compartments and colocalizes with actin filament bundles in the ellipsoid and calycal processes (arrows). Fluorescence is also bright in the axonal synapse at the outer plexiform layer (OPL). (B) The inner segment actin filament bundles (arrows) that extend into the calycal processes near the proximal outer segment (OS) are also labeled with Texas-Red phalloidin. A similar pattern of transgenic protein localization was found in rods containing serine 39 mutations to an alanine (C) or an aspartic acid (E). Fluorescence was predominantly in the rod cytosolic compartments, including the connecting cilium that can be seen in rods of the GFP/XF2S39D retina (E, arrowheads). However, some fluorescence colocalizing with actin filament bundles visualized with phalloidin (D, F) was also observed. OLM, outer limiting membrane. Scale bar: 10 μm.
Figure 4.
 
Localization of wild-type and mutant fascin 2 proteins tagged with green fluorescent protein (GFP) in transgenic Xenopus tadpole rods. (A) In transgenic Xenopus tadpole retina, GFP-tagged XF2 is abundant in rod cytosolic compartments and colocalizes with actin filament bundles in the ellipsoid and calycal processes (arrows). Fluorescence is also bright in the axonal synapse at the outer plexiform layer (OPL). (B) The inner segment actin filament bundles (arrows) that extend into the calycal processes near the proximal outer segment (OS) are also labeled with Texas-Red phalloidin. A similar pattern of transgenic protein localization was found in rods containing serine 39 mutations to an alanine (C) or an aspartic acid (E). Fluorescence was predominantly in the rod cytosolic compartments, including the connecting cilium that can be seen in rods of the GFP/XF2S39D retina (E, arrowheads). However, some fluorescence colocalizing with actin filament bundles visualized with phalloidin (D, F) was also observed. OLM, outer limiting membrane. Scale bar: 10 μm.
Figure 5.
 
Zebrafish fascin 2A and 2B bind actin filaments. (A) Purified GST/DrF2B fusion proteins sediment with actin filaments in the high-speed pellet (hsp). Mutation of serine 39 to an aspartic acid reduces the amount of fusion protein in the pellet. In the absence of actin, little of the fusion proteins is found in the high-speed pellet. GST alone does not bind to actin and is not found in the high-speed pellet in the absence or presence of actin. (B) A greater amount of the zebrafish fascin 2B fusion protein sediments with F-actin in the high-speed pellet compared to zebrafish fascin 2A. Fascin 2A fusion protein exhibits additional low-molecular–weight bands in the high-speed supernatant (hss), possibly indicating the presence of proteolytic fragments that do not bind actin filaments efficiently.
Figure 5.
 
Zebrafish fascin 2A and 2B bind actin filaments. (A) Purified GST/DrF2B fusion proteins sediment with actin filaments in the high-speed pellet (hsp). Mutation of serine 39 to an aspartic acid reduces the amount of fusion protein in the pellet. In the absence of actin, little of the fusion proteins is found in the high-speed pellet. GST alone does not bind to actin and is not found in the high-speed pellet in the absence or presence of actin. (B) A greater amount of the zebrafish fascin 2B fusion protein sediments with F-actin in the high-speed pellet compared to zebrafish fascin 2A. Fascin 2A fusion protein exhibits additional low-molecular–weight bands in the high-speed supernatant (hss), possibly indicating the presence of proteolytic fragments that do not bind actin filaments efficiently.
Figure 6.
 
Actin filament-binding activities of zebrafish fascin 2A and 2B and a S39D mutant of fascin 2B. Polymerized nonmuscle actin (7.1 μM) was combined with increasing concentrations of GST fusion proteins of zebrafish DrF2A, DrF2B, and the DrF2B serine 39 mutation to an aspartic acid. The amount of fusion protein associated with actin filaments relative to the amount of polymerized actin was determined by analyzing samples fractionated by centrifugation and visualized by SDS-PAGE. DrF2B had the greatest affinity for actin filaments with a significant loss of actin-binding activity when the conserved serine 39 was mutated to an aspartic acid to mimic a phosphorylated residue. The DrF2BS39D mutant does retain some residual low level of actin filament binding. DrF2A has a lower affinity for actin filament-binding compared with DrF2B but was also more insoluble during purification. However, DrF2A still has a greater affinity for actin filaments than the DrF2BS39D mutant. Error bars represent SD from at least four different samples.
Figure 6.
 
Actin filament-binding activities of zebrafish fascin 2A and 2B and a S39D mutant of fascin 2B. Polymerized nonmuscle actin (7.1 μM) was combined with increasing concentrations of GST fusion proteins of zebrafish DrF2A, DrF2B, and the DrF2B serine 39 mutation to an aspartic acid. The amount of fusion protein associated with actin filaments relative to the amount of polymerized actin was determined by analyzing samples fractionated by centrifugation and visualized by SDS-PAGE. DrF2B had the greatest affinity for actin filaments with a significant loss of actin-binding activity when the conserved serine 39 was mutated to an aspartic acid to mimic a phosphorylated residue. The DrF2BS39D mutant does retain some residual low level of actin filament binding. DrF2A has a lower affinity for actin filament-binding compared with DrF2B but was also more insoluble during purification. However, DrF2A still has a greater affinity for actin filaments than the DrF2BS39D mutant. Error bars represent SD from at least four different samples.
Figure 7.
 
A zebrafish GST–fascin 2B fusion protein bundles actin filaments. (A) When spun at low speed, filamentous actin remains in the supernatant (lss). If DrF2B fusion protein is added, actin filament bundles sediment into the low-speed pellet (lsp), and more zebrafish fascin 2B fusion protein sediments into the pellet than when the fusion protein is sedimented alone. GST alone does not bundle actin; the amount of actin filaments in the low-speed pellet remains unchanged in the presence or absence of GST. (B) If serine 39 is mutated to an aspartic acid residue in zebrafish fascin 2B fusion protein, sedimentation of actin filaments is lost. (C) Zebrafish fascin 2A fusion protein has dramatically less actin-bundling activity than zebrafish fascin 2B fusion protein. No change in actin filament sedimentation is observed when zebrafish fascin 2A fusion protein is added.
Figure 7.
 
A zebrafish GST–fascin 2B fusion protein bundles actin filaments. (A) When spun at low speed, filamentous actin remains in the supernatant (lss). If DrF2B fusion protein is added, actin filament bundles sediment into the low-speed pellet (lsp), and more zebrafish fascin 2B fusion protein sediments into the pellet than when the fusion protein is sedimented alone. GST alone does not bundle actin; the amount of actin filaments in the low-speed pellet remains unchanged in the presence or absence of GST. (B) If serine 39 is mutated to an aspartic acid residue in zebrafish fascin 2B fusion protein, sedimentation of actin filaments is lost. (C) Zebrafish fascin 2A fusion protein has dramatically less actin-bundling activity than zebrafish fascin 2B fusion protein. No change in actin filament sedimentation is observed when zebrafish fascin 2A fusion protein is added.
Figure 8.
 
Visualization of actin filament bundles cross-linked by zebrafish fascin 2B using fluorescence-tagged phalloidin. (A) Filamentous actin labeled with Alexa Fluor 488 phalloidin. (B) Adding 4 μM of a GST fusion to zebrafish fascin 2B resulted in a meshwork of thick, bundled actin filaments. (C) GST (4 μM) did not bundle actin filaments, though there might have been some protein aggregation. Scale bar: 5 μm.
Figure 8.
 
Visualization of actin filament bundles cross-linked by zebrafish fascin 2B using fluorescence-tagged phalloidin. (A) Filamentous actin labeled with Alexa Fluor 488 phalloidin. (B) Adding 4 μM of a GST fusion to zebrafish fascin 2B resulted in a meshwork of thick, bundled actin filaments. (C) GST (4 μM) did not bundle actin filaments, though there might have been some protein aggregation. Scale bar: 5 μm.
Figure 9.
 
Analysis of zebrafish fascin 2 actin filament-bundling activity. Increasing concentrations of GST/DrF2A and GST/DrF2B were incubated with 7.1 μM polymerized actin, and samples were analyzed for actin-bundling activity by low-speed centrifugation and SDS-PAGE. With increasing concentrations of GST/DrF2B, greater percentages of actin-filaments were observed in bundles found in the low-speed pellet. DrF2A showed bundling activity only at the highest concentrations tested. Error bars represent the SD from four or more samples.
Figure 9.
 
Analysis of zebrafish fascin 2 actin filament-bundling activity. Increasing concentrations of GST/DrF2A and GST/DrF2B were incubated with 7.1 μM polymerized actin, and samples were analyzed for actin-bundling activity by low-speed centrifugation and SDS-PAGE. With increasing concentrations of GST/DrF2B, greater percentages of actin-filaments were observed in bundles found in the low-speed pellet. DrF2A showed bundling activity only at the highest concentrations tested. Error bars represent the SD from four or more samples.
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×