The Actin-Binding Protein Capulet Genetically Interacts with the Microtubule Motor Kinesin to Maintain Neuronal Dendrite Homeostasis

Background Neurons require precise cytoskeletal regulation within neurites, containing microtubule tracks for cargo transport in axons and dendrites or within synapses containing organized actin. Due to the unique architecture and specialized function of neurons, neurons are particularly susceptible to perturbation of the cytoskeleton. Numerous actin-binding proteins help maintain proper cytoskeletal regulation. Methodology/Principal Findings From a Drosophila forward genetic screen, we identified a mutation in capulet-encoding a conserved actin-binding protein-that causes abnormal aggregates of actin within dendrites. Through interaction studies, we demonstrate that simultaneous genetic inactivation of capulet and kinesin heavy chain, a microtubule motor protein, produces elongate cofilin-actin rods within dendrites but not axons. These rods resemble actin-rich structures induced in both mammalian neurodegenerative and Drosophila Alzheimer's models, but have not previously been identified by loss of function mutations in vivo. We further demonstrate that mitochondria, which are transported by Kinesin, have impaired distribution along dendrites in a capulet mutant. While Capulet and Cofilin may biochemically cooperate in certain circumstances, in neuronal dendrites they genetically antagonize each other. Conclusions/Significance The present study is the first molecularly defined loss of function demonstration of actin-cofilin rods in vivo. This study suggests that simultaneous, seemingly minor perturbations in neuronal dendrites can synergize producing severe abnormalities affecting actin, microtubules and mitochondria/energy availability in dendrites. Additionally, as >90% of Alzheimer's and Parkinson's cases are sporadic this study suggests mechanisms by which multiple mutations together may contribute to neurodegeneration instead of reliance on single mutations to produce disease.


Introduction
Actin is a major component of the cellular cytoskeleton and one of the most abundant cellular proteins. However, actin must be exquisitely regulated during cell migration, cell adhesion, cell division, and many other essential cellular functions. Actin also forms the core of many cellular structures including filopodia, lamellipodia, microvilli and stress fibers. Actin exists predominantly in one of two forms: monomeric actin (G-actin) and filamentous actin (F-actin). The inter-conversion between these two actin forms is tightly regulated by a diverse array of proteins that bind actin directly or indirectly. Actin depolymerizing factor (ADF), also known as Cofilin, represents one actin-binding protein that can disassemble actin by severing and depolymerizing actin filaments (reviewed by [1,2,3,4]) However, actin severing can also contribute to actin assembly by generating actin fragments with exposed ends to seed new actin polymerization [5].
Dephosphorylation of Cofilin increases its activity, and Cofilin can bind both G-and F-actin (reviewed by [6]). Cofilin binding to F-actin, however, changes the filament conformation eliminating the phalloidin-binding site making the filaments phalloidin negative [7]. In addition, high levels of actin-bound Cofilin can also self assemble into aggregates [8,9,10,11]. The totality of experiments suggests that at any given moment, net actin polymerization and status depends not simply on actin levels, but on the balance of many distinct actin-regulatory factors and their activation status.
Another actin-binding protein that works in conjunction with Cofilin is Cyclase-Associated Protein (CAP), which exists in all eukaryotes [12]. CAP (also known as SRV2 in yeast) recycles Cofilin during actin depolymerization by displacing Cofilin from Cofilinbound G-actin, thereby freeing Cofilin for another round of actin binding and depolymerization [13,14]. Conversely, CAP may also function during actin polymerization by transferring G-actin to Profilin, which in turn adds the G-actin to the growing end of F-actin [15,16,17]. Previously, Capulet (Capt), the Drosophila CAP orthologue, has been shown to be involved in a number of developmental processes including oocyte polarity [18], cell morphogenesis [19], bristle morphogenesis [20], compound eye development [20] and wing disc dorsal/ventral pattern formation [19]. Capulet also has specific roles in neuronal function in Drosophila as it works directly with Abl to control axon guidance [21].
In neurons, extremely long neurites filled with G-actin must regulate the formation of F-actin in response to dynamic events such as synapse formation or axon guidance during sensation of chemo-attractive/chemo-repulsive cues. In addition, formation of ectopic F-actin must be suppressed to avoid physical blockages that could impede important transport functions within relatively thin neurites and produce deleterious cellular effects. For example, neurites contain a microtubule-rich cytoskeleton that provides a physical scaffold for delivery-both in anterograde and retrograde directions-for cargoes required to maintain proper neuronal function. Energy-dependent molecular motors, including dyneins and kinesins are ATPases that physically help deliver targeted cargoes by directional movement along these microtubules. In particular the kinesin superfamily protein KIF5 can transport diverse cargoes including membranous organelles, cytoskeletal proteins, and mRNAs (reviewed by [22]).
Forward genetic screens in model organisms provide one potential method for identifying molecules that regulate actin dynamics in vivo. Drosophila forward genetics has been applied to identify molecules required for processes ranging from embryo patterning to axon guidance. From a previous genetic screen, we identified a particular mutation that affected the accumulation of an actin, GFP fusion protein (actin::GFP), intriguingly only within neuronal dendrites but not axons [23]. We demonstrate here that loss of function mutations in capulet, are responsible for numerous actin-related phenotypes. Further, loss of function mutations in different genes synergize to produce severe phenotypes not observed by single loss of function mutations. In particular, we show that capulet genetically interacts with kinesin heavy chain and affects kinesin-mediated transport in the dendrites including mitochondrial distribution. Remarkably, the double mutant capulet and khc phenotype includes the formation of actin rods similar to those described in a Drosophila Alzheimer's model and in mammalian tissue culture neurodegeneration models. This study suggests a mechanism by which robust phenotypes may result from multiple seemingly mild impairments that interact with each other.

Identification of the capulet phenotype in dendrites
We previously conducted a forward genetic screen to isolate mutations affecting neuronal dendrite development [23]. One particular mutant, previously named ''punctate'', displayed actin::GFP-rich accumulations within dendrites but not within axonal shafts. However, neuron numbers and gross aspects of neural development including dendritic filopodia formation and dendritic branching appeared relatively normal within this mutant. These actin::GFP puncta, which are absent in wild type larvae, occur at an average of 38 puncta (standard deviation = 10) in each sensory neuron dorsal cluster of the punctate mutant. To determine the molecular lesion responsible for this phenotype, we used deficiency mapping and identified two molecularly defined deficiencies, Df(2L)Exel7006 and Df(2l)Exel6004, that failed to rescue the lethality of the punctate mutant. These two deficiencies overlapped by only 36 nucleotides within the Drosophila capulet (capt) gene. The Capulet protein encodes 424 amino acids with 3 recognizable motifs, namely a polyproline domain, a WH2 domain and two CARP domains. To confirm that capt was the affected gene, we sequenced the capt locus from the punctate mutant and found a nonsense mutation that changed amino acid lysine 304 to a premature stop codon. This mutant allele of capt, termed capt K304 , eliminated the last 121 amino acids of the Capulet protein, which includes both CARP domains. CARP domains are secondary structure predicted homology, not amino acid similarity, found in both Capulet and a gene mutated in Retinitis Pigmentosa [24]. The CARP domains bind to monomeric actin, while the polyproline stretch and WH2 domain also mediate interaction with the actin cytoskeleton [25,26,27,28].
To confirm that the dendrite actin::GFP phenotype was caused by mutations in capt, we determined whether deficiencies that uncover the capt locus would also display the actin::GFP phenotype seen in capt K304 ( Figure 1B). We successfully recombined the actin::GFP transgene onto Df(2L)ast5, a deficiency that lacks the capt locus, which also fails to complement the lethality of capt K304 . Indeed, we observed the actin::GFP phenotype when the deficiency was transheterozygous with capt K304 ( Figure 1C). We subsequently obtained a previously published allele of capt that encoded a nonsense mutation at tryptophan 145 (capt W145 ), and introduced the actin::GFP transgene into this mutant background. The capt W145 mutant also displayed the same actin::GFP phenotype when either homozygous or transheterozygous with capt K304 (Figure 1D-E). Similar phenotypes were observed with two additional P element insertion alleles capt k01217 and capt 06955 (data not shown). In addition to demonstrating nearly identical phenotypes with different alleles, we tested whether expression of a capt transgene in a capt K304 background would be sufficient to rescue the actin::GFP phenotype. To accomplish this, we made a fluorescently-tagged capt construct (UAS-mCherry::capulet) and recombined it onto the capt K304 chromosome. The capt K304 , UAS-mCherry::capulet recombinant did not display any actin phenotype ( Figure 1F), thus confirming the phenotype is due to the capulet mutation.

Capulet function in neurons
In yeast and Drosophila epithelial cells, capulet phenotypes are most obvious at cortical actin patches [29] and adherens junctions [30] respectively. Neurons, unlike yeast and epithelial cells, have more structurally isolated compartments due to their complex morphology. To examine the subcellular localization of Capulet in PNS neurons, we performed antibody immunostaining and expressed the UAS-mCherry::capulet transgene (UAS-mCC). In both cases, Capulet localized within cytoplasmic compartments including axons, cell bodies and dendritic shafts but did not localize to the Factin-rich dendritic filopodia [31] (Figure 2A). This is consistent with a role for Capulet as an actin monomer-binding protein since filopodia are primarily F-actin structures [32,33]. Capulet is thought to function by inhibiting inappropriate actin polymerization within cells as demonstrated by mutations in Drosophila, whether in epithelial cells [30] or eye cells [20] that demonstrate increased Factin in loss of function capt mutant cells. The converse hypothesis suggests that elevated capt function might decrease F-actin formation. We therefore tested this putative role on an F-actin-rich cellular structure, dendritic filopodia, by overexpressing wild type Capulet. We found that overexpression did suppress formation of F-actinrich dendritic filopodia ( Figure 2C), consistent with a role for Capulet as a potential inhibitor of filamentous actin formation and G-actin monomer-binding protein.
Actin-binding proteins can directly or indirectly influence the balance between free G-actin monomers and polymerized F-actin. We wondered whether the actin::GFP accumulations were stable or transient in nature. We observed that there did not appear to be a substantial net increase in actin::GFP-rich aggregates in older capt K304 larvae compared to younger capt K304 larvae, suggesting that actin::GFP puncta did not accumulate over time. To directly assess the dynamics of the actin::GFP phenotype we performed time lapse imaging of actin::GFP expressing capt mutant larvae. Time lapse analysis revealed that the actin::GFP phenotype is dynamic with accumulations dissipating within dendrites, while simultaneously new accumulations can form within a different part of the same dendrite ( Figure 2D-F). This suggests the possibility that Capulet functions to regulate the balance between actin in filaments and actin in soluble monomeric form within dendrites.

Capulet and cofilin interaction in neurons
Previous studies demonstrate that Capulet interacts with Cofilin to promote actin turnover and that the absence of Capulet can lead to formation of Cofilin/actin aggregates [34]. We utilized genetic analysis to examine the genetic relationships between capulet and cofilin. Overexpression of Cofilin in mammalian cells has been demonstrated to lead to the formation of ADF/Cofilin (AC)-actin rods [35]. These investigators proposed that high levels of actin bound Cofilin can lead to self-assembly and actin-rod or actinpuncta formation [35]. Upon over-expression of Cofilin in wild type da neurons we observed formation of actin::GFP accumulations with occasional rod-like structures forming ( Figure 3B). However, these actin-rods appear to be greater in number and longer when cofilin is overexpressed within a capt K304 background ( Figure 3C), suggesting that capulet helps reduce actin-rod formation by interacting genetically with cofilin. Conversely, no actin rods form when cofilin and capt are simultaneously over-expressed ( Figure 3D), suggesting that capulet and cofilin genetically antagonize each other. These results, however, are also consistent with the biochemical models in which Capulet helps recycle Cofilin and enhance it's depolymerization activity [13,14,34].
In mammalian cells, AC-actin rods do not usually stain with phalloidin but are instead decorated with Cofilin [36]. We phalloidin stained the capulet, khc double mutant and immunostained for Cofilin. Both the actin puncta and the actin rods failed to stain with phalloidin yet both contained abundant Cofilin ( Figure 3E-F), suggesting they share some features with AC-actin rods described in mammals.

Capulet mutations alter mitochondria distribution in dendrites
Depletion of ATP in mammalian cells has been shown to generate AC-actin rods. Further, in mammalian neurites the ACactin rods fill the diameter of neurites essentially blocking transport within the neurite and severely disrupting the microtubule network (reviewed by [37]). Due to the long, thin architecture of sensory dendrites, it is possible that the formation of the Cofilin-decorated actin puncta in the capt K304 mutant could also hinder transport within these dendrites. Mitochondria are relatively large organelles that are actively transported along microtubule tracks within dendrites to provide ATP throughout the dendritic tree. We determined whether the transport of mitochondria within dendrites is affected in capt K304 by expressing mito-GFP, a mitochondrial targeting sequence fused to GFP, in the PNS to visualize the distribution of mitochondria. In a capt K304 /+ heterozygote, the mitochondria are dispersed throughout the length of the dendrites, whereas in the capt K304 homozygote, there are significantly fewer mitochondria dispersed in the dendrites with approximately a 50% reduction in dendritic numbers ( Figure 4A-C). However, the distribution of mitochondria in the cell bodies and in the axons between the capt heterozygote and capt homozygote are comparable. This suggests that Capulet could help maintain dendrites clear of actin polymerization that may otherwise impede microtubule-based transport such as mitochondria transport.

Capulet genetically interacts with kinesin heavy chain
We wondered what genes, other than actin-binding proteins, might be relevant to capulet function, which could also explain the mitochondrial phenotype. Previous studies in Drosophila have demonstrated that mutations in capulet and kinesin heavy chain (khc) had similar phenotypes in establishing anterior-posterior asymmetry. The anterior localization of bicoid mRNA and posterior localization of oskar mRNA are essential in organizing the anterior- posterior asymmetry in Drosophila oocytes. One study documented that a mutation in capt alters the posterior localization of oskar mRNA [18]. In a separate study, a mutation in kinesin heavy chain (khc) was also shown to perturb oskar mRNA distribution [38]. In addition, the major mediator of transport of mitochondria in axons and dendrites is Kinesin heavy chain (Khc) [39]. Since both capt and khc mutations produced the same oskar mRNA mislocalization phenotypes in oocytes and Khc transports mitochondria, we investigated whether there was any genetic interaction between capt and khc. To determine whether genetic interactions exist between capt and khc, we placed an actin::GFP transgene in a khc mutant (khc k13314 ) background. Homozygous khc k13314 mutants exhibited swellings in dendrites ( Figure 5A), similar to the axonal swellings previously observed in khc mutants [40], whereas these swellings were absent in a khc k13314 heterozygote. When we crossed the capt K304 line with the khc k13314 line, the capt, khc transheterozygote showed dendritic swellings, confirming that these two genes genetically interact in a doubly heterozygous animal ( Figure 5B). In a capt, khc double loss of function mutant, both the actin::GFP puncta and dendritic swellings, seen in capt and khc single mutants respectively, occurred at higher frequency ( Figure 5C), suggesting that they enhance each other. Strikingly, however, we also observed the formation of elongate actin rods in the double mutant ( Figure 5C), not seen in either single mutant. Actin::GFP puncta in capt mutants had an average length of 1.23 mm (standard deviation = 0.21 mm), whereas the actin::GFP rods we observed in the capt, khc double mutant had an average length of 5.97 mm (standard devia-tion = 1.75 mm). The average number of actin rods (.2 mm in length) found in the dorsal cluster sensory da neurons was ,7 per cluster (mean = 7.33, standard deviation = 2.77). Importantly, actin rods were never seen in wild type larvae. We tested the genetic interactions between capt and khc using a different allele of capt (capt W145 ) or a different allele of khc (khc e02141 ) and we were able to replicate the same findings (data not shown).
Alteration of kinesin function is associated with swellings in the plasma membrane in axons [40,41,42]}. We visualized the dendrite membrane by expressing myristoylated-mRFP (red fluorescent protein), which targets to the plasma membrane, to determine whether or not the actin::GFP accumulations affect the diameter of the surrounding dendrite. Surprisingly, unlike kinesin mutations that are associated with membrane swellings, neither the actin puncta nor the actin rod-like structures are associated with localized swellings. (Figure 5D-I).
To compare the neuronal distribution of Khc in a WT and capt background, we expressed a Khc::EGFP transgene in the PNS neurons. We could not easily visualize Khc::GFP along the dendrites of PNS neurons in both WT and capt backgrounds with a single overexpressing transgene. However, Khc::EGFP appears to accumulate in the cell bodies in a capt heterozygote which is not observed in a WT background (Figure 6 A-B), suggesting that Khc::EGFP movement from cell body to dendrites is disrupted when Capt levels are reduced to half. We attempted to make a recombinant to express more Khc::GFP to more easily visualize dendrites, however, we discovered that the capt, Gal4 109(2)80; UAS-Khc::GFP transheterozygote is synthetic lethal and dies at the

Discussion
Regulation of actin dynamics has important consequences in all cell types but special significance in neurons as dendritic spinesthe sites of synapse formation in the brain -and axon growth cones are highly enriched for F-actin. In addition, neurites, either axons or dendrites, span very long distances from the cell body and therefore must allow continued transport of proteins to and from the cell body to either axon termini or dendritic spines. Blockage of relatively small diameter dendrites could hinder this transport and presumably have deleterious affects on neuronal function.
Capulet plays a vital role in regulating actin dynamics and can reduce filament formation by binding/sequestering actin monomers and preventing new polymerization or by recycling Cofilin activity by removing actin monomers from it, thus freeing Cofilin for another round of actin severing and depolymerization [43]. Consistent with this function, multiple screens in Drosophila have identified capulet loss of function mutations that lead to ectopic increased F-actin accumulation [20,30]. When Capulet function is reduced, endogenous Cofilin may have less F-actin depolymerizing activity leading to actin filament accumulation. Alternatively, very high levels of Cofilin, generated by over-expression experiments in mammalian cells or in this study with transgenic animals, can exceed the capacity of Capulet to recycle Cofilin depolymerizing activity, leading to self-assembly of actin-bound Cofilin into AC rods. In essence, the balance between Capulet and Cofilin levels and their activity determines the net actin polymerization status. Our genetic studies support various biochemical and tissue culture models whereby decreases in capulet function lead to increased actin in small aggregates, while over-expressed Cofilin in a capulet mutant background leads to larger accumulation in actin rod-like structures.   In our study, Capulet maintains dendrites clear of ectopic actin filaments that otherwise could physically disrupt microtubulebased function, including Kinesin-mediated transport. In addition to other cargoes, Kinesin also transports mitochondria in the anterograde (plus-end) direction along microtubules [38,45]. Impaired Kinesin function could then diminish mitochondrial transport, which are transported hundreds of microns from the cell body in dendrites, leading to less available ATP. The transheterozygous genetic interaction we observe between capulet and khc as well as the robust double mutant phenotype that produces actin rods in dendrites suggests a strong cellular interdependence between these two molecules (see model Figure 7).
A fundamental question relates to why we only observe dendrite and not axon phenotypes. It is highly unlikely that capulet only functions in dendrites, as roles for capulet in actin regulation have already been demonstrated in diverse cell types from follicle epithelial cells to the adult compound eye [20,30]. The simplest interpretation is that the abundance of actin-based structures in these dendrites, dendritic filopodia, which are not found on axons, represents a larger pool of actin to be regulated. Alternatively, the internal diameters of these dendrites are much smaller than their axons suggesting any impairment of cytoplasmic flow -by diffusion or transport-would be preferentially obstructed in the dendrites. Indeed, the distribution of mitochondria in cell bodies and axons are normal in capulet mutants, however, relatively smaller Drosophila sensory dendrites would be more susceptible to impaired transport and decreased ATP.

The role of actin in neurodegenerative disease
The role of altered actin polymerization and aggregation in neurodegenerative disease is relatively unexplored, although some studies have begun to examine actin aggregates in neurodegenerative disease. In Alzheimer's disease brain, aggregates of ADF/ Cofilin have been described in association with both amyloid deposits and neurofibrillary tangles [36,46]}. There are also aggregates of ADF/Cofilin in actin-crystalline structures termed Hirano bodies, which have been documented in distinct neurodegenerative diseases including Alzheimer's and Parkinsons [47]. One might speculate that alterations in actin dynamics may not be a causative factor in neurodegenerative conditions but perhaps a contributing factor to neuronal dysfunction.
Recently, investigators have described an Alzheimer's disease model developed in the Drosophila brain with altered actin dynamics. In this study, the authors documented that ectopic expression of a mutant human Tau protein, associated with fronto-temporal dementia, increases the levels of F-actin in the brain [48]}. Further, they observed actin rod structures in the brains of both the Drosophila model and in a mouse transgenic tauopathy model. Some of the rods stained for actin, Cofilin and Tau, which are all characteristic of Hirano bodies. It is interesting to note that in their study, co-expression of Cofilin with the mutant Tau protein significantly reduced the effects of expressing mutant Tau alone, suggesting that they genetically antagonize each other. Conversely, over-expression of human wild type Tau enhances the toxicity of b-amyloid (Ab). These authors suggest that interactions between the microtubule binding protein Tau and actin may be a key mediator of Tau-related neurotoxicity [48]. We did not see a genetic interaction in a heterozygous capulet background when expressing the human mutant Tau protein [48] (data not shown). Our results suggest mechanisms by which loss of capulet function interacts with impaired Kinesin function to produce similar actin rods. Interestingly, both studies suggest that proteins normally associated with microtubules can genetically interact with the actin cytoskeleton under pathological conditions. In many neurodegenerative models impaired Kinesin function and the failure of motor-driven transport in long axonal compartments is hypothesized to play a major role in neurodegeneration. This transport failure leads to formation of axonal swellings, which are the physical manifestation of accumulation of axonal cargoes. For example, overexpression of Drosophila bAPP results in axonal transport defects, which are enhanced by genetic reductions in the amount of functional Kinesin [49]. bAPP and its proteolytic machinery undergo Kinesin 1-mediated fast anterograde axonal transport in neurons [50]. Interestingly, mutations in Tau, a microtubule binding protein, are also extensively documented in Alzheimer's (reviewed in [51,52]).

Are actin rods protective?
Although actin rods are induced under conditions that diminish neuronal function or with expression of neurodegenerationassociated proteins, at least one study postulates that actin rod formation may be a neuroprotective response [35]}. Neuronal cell culture models designed to mimic neurodegenerative insults, including excitotoxicity and oxidative stress, induce formation of ADF/Cofilin-actin rods within neurites. When examined by electron micrographs these ADF/Cofilin-actin rods physically displace microtubules from the neurite by filling it [36]. While these rods would be clearly detrimental to neural function in the short term, in the long term the authors speculate that by possibly inhibiting actin treadmilling the rods may decrease ATP consumption. As an ATPase, active actin treadmilling consumes ATP, which could be problematic under cellular stress. Both the small actin aggregates documented here (by time-lapse analysis), and the actin rods documented in mammalian cells can be dynamic and dissipate and return to normal actin pools, allowing for the possibility that they could be transiently neuroprotective. If the actin rods were neuroprotective they would have to be relatively short-lived. In our investigations, capt/khc double mutant larvae do not survive long enough under time lapse to observe the disappearance of actin-rods (data not shown).
In addition to forming actin rods by overexpressing Cofilin or microinjecting Cofilin, rod formation can also be induced simply by depleting ATP. Decreasing ATP would alter numerous cellular pathways. However, Cofilin has a higher affinity for ADP/actin than ATP/actin and by self-assembling ADP/actin/Cofilin into rods, the exchangeable ADP/actin could be removed from the dynamic actin pool decreasing ATP turnover. Interestingly, actin rods are not unique to neurons, as non-neurons under stress can also be induced to form Cofilin positive, phalloidin negative actin rods [53]. Reorganization of actin stress fibers (that stain with phalloidin but not Cofilin) into cytoplasmic and intranuclear actin rods decorated with Cofilin but not phalloidin have also been produced in mouse fibroblasts by exposing the cells to stressors [53].

Other actin regulators and kinesin function
A second, line of thought suggests that other actin regulatory proteins may directly interact with kinesin and could provide another route for Capulet ultimately affecting Kinesin-mediated transport. Enabled (Ena) is an actin-polymerizing factor that genetically antagonizes the Abl tyrosine kinase (reviewed by [54]). Reduced Ena function has been demonstrated to rescue khc phenotypes-paralytic tail flipping and axonal swellings -in a sensitized Abl background [55]. These investigators suggest that Ena normally inhibits kinesin-mediated transport by a direct physical association. However, as Ena facilitates actin polymerization, an alternative interpretation could be that unchecked by Abl, excessive Ena activity could lead to ectopic actin polymerization, which could hinder kinesin-mediated transport. Abl, which inhibits Ena activity, has a SH3 domain that could interact directly with Capulet's polyproline domain and Abl and Capulet have already been demonstrated to co-immunoprecipitate [21]. In fact, previously it was demonstrated that Abl and Capulet serve a common function in ultimately promoting repulsion of CNS axons to the midline [21]. It would be consistent for capulet and abl to also have a common function in promoting kinesin movement in microtubules, as both of these mutants form swellings in neuritic processes when transheterozygous with khc. Furthermore, both Abl and Capulet prevent formation of F-actin, either by Capulet depolymerizing F-actin via Cofilin, or by Abl negatively regulating Ena.
In summary, we demonstrate a strong genetic interaction between kinesin and capulet. A capulet-khc transheterozygote forms dendritic swellings, similar to the dendritic swellings seen in a khc mutant alone. However, neither reduced function of either capulet nor kinesin results in actin rods. Only the double mutant of both produces AC-like actin rods observed in numerous animal and cell culture models of neurodegeneration. Taken together, our observations suggest that capulet is necessary for kinesin-mediated dendritic transport and in turn cargoes transported by kinesin, like mitochondria, have vital roles in regulating actin dynamics in dendrites. While an appreciation of actin dysregulation as a potential contributor to neurodegenerative disease is just beginning, this study illustrates the severe consequences of altered actin regulation in combination with other defects. In this sense, it is interesting to note that .90% of Alzheimer's and Parkinson's cases are sporadic [36,56,57], suggesting that perhaps multiple defects in different proteins/genes may combine to cause sporadic neurodegenerative diseases.
To the best of our knowledge, this is the first case of Cofilindecorated actin rod formation due to defined genetic loss of function mutations. Previous actin rods have been generated by subjecting cells to stress (ATP depletion, heat shock treatment, oxidative stress by hydrogen peroxide treatment, glutamateinduced excitotoxicity, and DMSO treatment) or by overexpressing proteins such as Cofilin, Tau and b-Amyloid peptide in transgenic models.

Fly stocks
We follow guidelines from Flybase (www.flybase.org) for nomenclature. Gene names are italicized, whereas proteins are not. To visualize PNS neurons, Gal4 109 (2)

Mapping and gene identification of punctate
Virgin females of punctate (punc/CyO) were crossed with males from each stock of the Chromosome 2 Deficiency Kit from Bloomington. We noted which deficiency stocks did not generate F1 progeny with straight wing flies (i.e., punc/def is lethal). We identified the common chromosomal locus among all the deletions in these stocks, which is within the capulet gene. To identify the nature of the EMS mutation, we amplified and sequenced overlapping segments of capt from genomic DNA of punctate and compared it to the WT capt sequence.

Cloning
To make the mCherry::capt (monomeric Cherry::Capulet) fusion construct, mCherry::Tubulin was taken out of pcDNA3.1 (gift from the R. Tsien lab) by cutting with EcoRI. Both ends of the excised mCherry sequence were made blunt by Klenow fragment (New England BioLabs, Ipswich, MA) treatment and the 39 end was made sticky by digestion with XhoI. The pUAST vector was linearized by digestion with HindIII and the ends were made blunt by Klenow fragment treatment. The 59 end of the pUAST vector was cut with XhoI to create a sticky end. The mCherry::Tubulin fragment was then inserted into the pUAST vector by blunt-sticky ligation. The capulet gene sequence was amplified from the cDNA clone LD24380 by using a forward primer containing a BglII restriction site and a reverse primer containing an XhoI restriction site. The capulet PCR product was digested with BglII and XhoI. The tubulin sequence was excised out of the newly constructed pUAST-mCherry::Tubulin by digestion with BglII and XhoI and was replaced with the capulet sequence. The pUAS-mCherry::capulet construct was transformed into electrocompetent DH10B bacteria, amplified and injected into w 1118 Drosophila embryos by P element mediated transformation to make transgenic fly lines (Duke Drosophila Transgenic Core).
Polyclonal antibodies against His-Capulet were generated by linearizing the pET28a vector with BamH1 and XhoI digestion and ligating the amplified capulet PCR product (see above) that had been digested with BglII and XhoI. The capulet in the pET28a vector was transformed into competent cells, amplified, sequence verified, and transformed into BL21 cells. The His-Capulet protein was extracted and purified using the TALON system and sent to QED Bioscience Inc. (San Diego, CA) for antibody production.

Immunohistochemistry
The dissection of larvae, immunostaining procedures and confocal microscopic visualizations were done essentially as described (Medina et al., 2006). Briefly, the larva was washed with water and placed in a droplet of 4% para-formaldehyde. The head and the tail sections of the larva were excised using a pair of fine scissors and the guts and other visceral organs were pulled out with forceps. An anterior-posterior cut was made along the middorsal area of the larva and the resulting larval fillet was cleaned out of muscle tissue with forceps. The larval fillet was fixed in 4% paraformaldehyde for 20 minutes and then washed with PBT 3 times. This was followed by incubation with a blocking solution of 5% normal goat serum and PBST for one hour at room temperature. The larval fillet was incubated with the primary antibody overnight at 4uC, washed 3 times with PBT, incubated with the secondary antibody for 4 hours at room temperature and then washed again 3 times with PBT. The larval fillet was mounted on a slide with glycerol to prepare for viewing. The primary antibodies used were anti-twinstar/cofilin (1:50; gift from J.R. Bamburg), anti-actin (1:200, A4700; Sigma, St Louis, MO), and anti-GFP (1:1000; Molecular Probes, Euegene, OR). Staining actin with phalloidin (0.16 mM, Alexa Fluor 546 phalloidin; Invitrogen, Carlsbad, CA) was performed according to manufacturer's instructions, and when double labeling, was done after washing off the secondary antibodies.

Image acquisition and processing
For live and fixed larva visualization of the PNS, images were obtained using a Zeiss LSM (Laser Scanning Microscope) 510 (Germany) confocal microscope with a 406 and a 636 oil immersion lens. For live specimens, larvae were picked from food bottles and vials, cleaned with water, mounted on a slide with halocarbon and covered with a cover slip. Fixed specimens were washed with 16 PBT and mounted on a slide with glycerol and covered with a cover slip. Database microscopic images taken using the LSM510 software exported as TIFF files, cropped and sized using Adobe Photoshop (San Jose, CA) and arranged and labeled using Adobe Illustrator (San Jose, CA).