Genetic Modifier Screens Reveal New Components that Interact with the Drosophila Dystroglycan-Dystrophin Complex

The Dystroglycan-Dystrophin (Dg-Dys) complex has a capacity to transmit information from the extracellular matrix to the cytoskeleton inside the cell. It is proposed that this interaction is under tight regulation; however the signaling/regulatory components of Dg-Dys complex remain elusive. Understanding the regulation of the complex is critical since defects in this complex cause muscular dystrophy in humans. To reveal new regulators of the Dg-Dys complex, we used a model organism Drosophila melanogaster and performed genetic interaction screens to identify modifiers of Dg and Dys mutants in Drosophila wing veins. These mutant screens revealed that the Dg-Dys complex interacts with genes involved in muscle function and components of Notch, TGF-β and EGFR signaling pathways. In addition, components of pathways that are required for cellular and/or axonal migration through cytoskeletal regulation, such as Semaphorin-Plexin, Frazzled-Netrin and Slit-Robo pathways show interactions with Dys and/or Dg. These data suggest that the Dg-Dys complex and the other pathways regulating extracellular information transfer to the cytoskeletal dynamics are more intercalated than previously thought.


Introduction
Muscular dystrophies are a group of inherited neuromuscular disorders that share the same basic phenotype of progressive loss of muscle integrity. Many muscular dystrophies are caused by defects in a specialized cell adhesion complex called the Dystrophin Glycoprotein Complex (DGC). It has become evident that this complex plays a central role in muscle integrity and forms a mechanical link from the actin cytoskeleton to the extracellular matrix (ECM). The core DGC is composed of a transmembrane component, Dystroglycan (Dg), which associates with the ECM protein, Laminin and the cytoplasmic protein Dystrophin which binds Actin (reviewed in [1,2]).
Many lines of evidence confirm that maintaining the structural link from the extracellular matrix to the actin cytoskeleton is crucial in preventing many forms of muscular dystrophy. Mutations that disrupt any component of this structural link results in a variety of muscular dystrophies like Duchenne's, Becker's, Muscle-eye-brain disease, Walker-Warburg syndrome, congenital muscular dystrophies 1C and 1D as well as limb girdle muscular dystrophy 2I. These diseases share the common symptoms of skeletal muscle degeneration, cardiomyopathy, as well as a reduced life span for afflicted individuals [3].
Additionally, alterations which reduce the affinity of components of the DGC lead to congenital muscular dystrophies like Fukuyama's which, aside from muscular defects, also are associated with aberrant neuronal migrations that lead to mental retardation, epilepsy, as well as abnormal eye development. The use of animal model systems has led to the clarification of the roles of specific gene products in maintaining muscle integrity and function (reviewed in [4]), however, the regulation of this complex is largely unknown.
Initial characterization of the DGC in Drosophila has determined that components studied so far possess similar roles in muscle integrity and neuronal migration in flies as in humans (Figure 1, [5,6,7,8]). These abnormalities include age dependent muscle degeneration, reduced mobility, defects in eye development as manifested by altered photoreceptor axon pathfinding, and a shorter life span. Additionally, mutations in Dys and Dg affect cell polarity in the Drosophila germ line as both the follicular cell epithelium and the oocyte are disrupted [6,9,10]. Recently, a reduced lifespan in Drosophila, as well as heart and muscle abnormalities, have been reported in mutants of another component of the DGC, d-sarcoglycan [5]. In addition, heart and eye phenotypes have been observed in Drosophila Dys and Dg mutants [11,12].
The similar defects in both flies and humans make Drosophila an attractive model for further studies on clarifying the role of the DGC. Such studies may reveal novel components that may likely have counterparts in humans. Additionally, since very little is known about how the DGC is regulated insights may be gained on this heretofore unknown process. More recently, transmembrane signaling has been implicated in the function of the DGC. The Cterminus of Dystroglycan, in addition to having, EF and WW domain binding sites, also possesses SH2 and SH3 domain binding sites. These known protein-protein interaction motifs support the idea that Dystroglycan is a signaling receptor in addition to its known role as a conduit between the ECM and the cytoskeleton. Changes in MAPK kinase and GTPase signaling have also been observed when the DGC is perturbed [2,13]. Recent work has shown that specific sets of domains are critical in the function of Drosophila Dystroglycan [8].
In the present work, we have used the genetic tractability of Drosophila to search for novel components of the DGC, as well as components that may be involved in its signaling and regulation. Such a search is straightforward because in addition to the muscle degeneration and photoreceptor axon pathfinding defects, mutations in dystrophin and Dystroglycan cause a visible phenotype manifested as alterations in the fly wing, particularly the posterior crossvein ( Figure 2, [14]). Since this is an easily score-able, highly penetrant phenotype we undertook a dominant modifier screen approach and looked for flies that showed either a suppression of the crossvein phenotype or a noticeably altered crossvein. Importantly, crossvein development has been previously shown to require EGFR, TGF-b and Notch pathway activities and is therefore a sensitive place to observe potential interactions of the DGC with these signaling pathways [15,16,17]. In addition, hemocyte migration is shown to correlate with the crossvein development [14]. Therefore genes involved in correct migration processes might also be obtained by this approach. We screened Pelement lethal as well as deficiency collections for interactors in addition to performing a classical ethylmethanesulfonate (EMS) screen for dominant modifiers.
Here we report that in using these screening strategies we have found modifiers that belong to six different functional groups. We have found genes involved in muscle development, neuronal/cell migration and motor function as well as cytoskeletal components and components of the TGF-b, EGFR and Notch pathways. A common theme among many of these interactors and Dg-Dys complex is their involvement in the cytoskeletal rearrangements controlled by extracellular cues.

Results
To identify genes that interact with Dys and Dg, we used the chromosomal lesion hypomorph Dys E6 as well as the RNAi knockdown mutants Dys N-RNAi , Dys C-RNAi and Dg RNAi . These mutants exhibit age dependent muscle degeneration of indirect flight muscles (IFM, Figure 1B and 1C) as shown before for other Dys alleles [6].
In addition to the muscle phenotype, a visible wing vein phenotype was observed in these and previously analyzed alleles. In particular, the posterior crossvein (PCV) is defective in both Dys and Dg mutants, a disrupted anterior crossvein (ACV) and a partial duplication of L2, a longitudinal wing vein, can also be observed at a lower frequency ( Figure 2). These phenotypes are present in the RNAi knockdown mutants, Dys N-RNAi , Dys C-RNAi and Dg RNAi ( Figure 2C, 2D and 2E, respectively) as well as the chromosomal lesion hypomorphs Dys E6 ( Figure 2B), DysDf, Dg O43 , Dg O55 and Dg O86 [14]. Dys and Dg are expressed in third instar larval wing imaginal discs ( Figure S1) and interact in the wing vein since the double mutant shows a new phenotype -a duplication of L3 longitudinal wing vein ( Figure 2F and 2F').
We used the visual wing vein phenotypes in P-element lethal, EMS induced and deficiency screens to find dystrophin and Dystroglycan interactors in Drosophila that either increase or decrease the wing vein phenotype. Additional studies reveal that many of the interactors identified in the screens are required for proper function in other tissues that require Dys and Dg, such as muscle, brain and ovary.

Modifier Screens
To identify Dys and Dg interactors, we screened three different kinds of mutants; a collection of 800 FRT-P-element lethal lines, a deficiency collection of 216 lines and 43,000 EMS induced mutants. From these analyses, we identified 37 interacting genes that can be clustered into six different functional groups (Table 1).
P-element screen. We screened FRT-P-element lethal lines from the Kyoto Stock Collection (Japan). The homozygous lethality of the lines allows us to infer that the transposon is inserted in or near an essential gene. We set up crosses to search for modifiers of the RNAi mutants Dys C-RNAi , Dys N-RNAi and Dg RNAi as well as Dys E6 , a hypomorph that removes DLP2, a specific long isoform of dystrophin [7]. Modifiers were divided into phenotypic classes of enhancers, suppressors and suppressors with extra wing vein material (En, Su or Su+; Figure 3). From the P-element screen we obtained 33 modifiers of the Dys phenotypes (Table 1), 25 of which also showed interaction with Dg ( Figure 4). Since the identity of these genes is known, we were able to organize these modifiers into 6 different functional categories (Table 1).
We were excited to find genes involved in both muscle and motor/cytoskeletal function as these are processes in which Dys and Dg are known to act (Table 1, functional group I; Figure 4). Of the muscle category genes, muscleblind (mbl) was the strongest suppressor in the wing vein. Weak suppressors of this group included Calmodulin (Cam) and the nicotinic acetylcholine receptor a 30D (nAcRa-30D) ( Table 1, functional group I), though nAcRa-30D strongly interacted with Dg ( Figure 4). Cytoskeletal and motor function genes (Lissencephaly-1 (Lis-1), Kinesin heavy chain (Khc), Dynamitin (Dmn) and Fhos showed moderate to weak suppression with dystrophin mutants (Table 1, functional group I).
Another intriguing functional group that interacts with Dg-Dys complex includes genes known to be involved in neuronal migration (Table 1, functional group II). This group is composed of the genes which belong to Semaphorin-Plexin (Sema-1a, Sema-2a), Slit-Robo (slit (sli), roundabout (robo), leak (lea) -a robo2 homolog, Syndecan (Sdc)) and Netrin-Frazzlez (frazzled (fra)) pathways. In the wing vein most of these genes were scored as moderate to weak suppressors of the Dys RNAi mutants. Importantly, Sema-2a and fra modify Dys E6 (Table 1, functional group II; Figure S3). Sema-1a was found to moderately enhance the Dys wing vein phenotype. Using our Dg RNAi mutant we found that robo and Sema-2a enhanced the Dg phenotype and Sema-1a and lea (robo2) suppressed the Dg phenotype (Table 1; Figure 4 and Figure S4). We have previously shown that Dys and Dg function in the Drosophila Photoreceptor axon pathfinding [6] in processes similar to Semaphorin-Plexin, Slit-Robo and Netrin-Frazzled pathways.
Functional groups IV and V ( Table 1) contain interactors that belong to the TGF-b signaling pathway: Daughters against dpp (Dad), decapentaplegic (dpp), thickveins (tkv) and EGFR signaling pathway: kekkon-1 (kek1) and argos. Dad, dpp and kek1 were strong modifiers of the Dg RNAi mutant (Table 1; Figure 4 and S4) and Dad and tkv of Dys E6 allele as well (Table 1, functional group IV; Figure S3). It has been known for some time that TGF-b, Notch and EGFR pathways are necessary for proper wing vein development ( [18]; reviewed in [19,20]). In fact, many novel factors of these pathways have been identified through the analysis of wing vein mutants. The longitudinal wing veins that begin to form during late larval stages of development require EFGR and TGF-b signaling pathways for proper fating of cells in the region [15,21]. Crossveins appear in late pupal stages of development and require TGF-b signaling for formation and EGFR and Notch signaling for final patterning [17]. The posterior crossvein is particularly sensitive to different levels of TGF-b signaling and forms only after the proper formation of the longitudinal veins.
The last functional group (Table 1, functional group VI) contains a group of 12 genes with disparate functions. One enhancer, glial cells missing (gcm), was identified. Three strong suppressors of this group also interacted with Dys E6 , the Drosophila homolog of the homeodomain-interacting protein kinase (HIPK), kismet (kis) and l(3)L4092, which contains a zinc-finger motiff. kis, deadlock (del), gcm, POSH and wunen (wun) also strongly modified Dg RNAi (Figure 4).
Heparan sulphate proteoglycan (interacts with Sli and Robo) Deficiency screen. In addition to the P-element lethal collection, we also screened for the Dys interactors using a collection of deficiency lines, which covers about 30% of the 1st, 40% of the 2nd and 80% of the 3rd chromosome and found 10 regions on the 2nd and the 3rd chromosomes that interacted with Dys N-RNAi in the wing vein ( Figure 5A). Nine of these deficiencies suppressed the Dys PCV phenotype with formation of extra wing vein material and belonged to the Su+ class ( Figure 3C and 5B) and one to the En class ( Figure 3D and 5B). The wg and stan (Flamingo, Fla) genes, which belong to the planar cell polarity pathway (Table 1), were identified as Dys N-RNAi interactors in [27D1-27F2] and [47A7-47C6] cytological regions respectively ( Figure 5B). Both of these genes modify Dg RNAi phenotype in wing vein as well ( Figure 4). A third component of the planar cell polarity pathway, grainy head (grh), was identified from the Pelement screen. Furthermore, dystrophin and Dl genes located in region [91F12-92A11], as well as poly-EGF and CG4496 from [27D1-27F2] region were also independently identified in the Pelement and EMS screens ( Figure 5B and Table 1).
EMS mutagenesis screen. We analyzed ,37,000 chromosomes for modifiers of Dys N-RNAi and Dys N2-RNAi and over 6000 chromosomes for enhancers of Dys E6 and isolated 27 modifiers (Table S1). Eighteen of these were localized to the 2nd chromosome and nine were localized to the 3rd chromosome. The genes defective in eight of these mutants were identified and shown to correspond to five genes ( Table 1). The modifier screens using EMS produced similar phenotypes to those observed in the P-element screen ( Figure 4) and additionally produced a ''posterior crossveinless'' class (14/27 modifiers; Figure S2). These modifiers may be alleles of the crossveinless genes on the second and third chromosomes. There are eight previously identified crossveinless loci, cv, cv-2, cv-3, cv-b, cv-c, cv-d, cvl-5 and cvl-6. Two, cv and cvl-6, are on the X chromosome. One, cv-3 is on the second chromosome and the remaining five are on the third chromosome. Modulators of the TGF-b pathway are encoded by two of the eight crossveinless loci, cv and cv-2. Of the remaining six one other, cv-c has been molecularly characterized and encodes a Drosophila Rho GTPase Activating Protein. The others may also encode effectors of TGF-b signaling. Nevertheless, when cv and cv-2 are lost the resultant aberrant TGF-b signaling results in the loss of both crossveins. Of our crossveinless like modifiers, one, Mod90 (Table 1 and Table S1), was mapped to cytological location 66B, two map units distal to the hairy gene, and so does not appear to be an allele of any of the crossveinless genes on the third chromosome. An attractive candidate for this gene is moleskin (msk) which encodes Drosophila Importin-7, a protein involved in nuclear translocation that has been shown to regulate the TGF-b pathway by controlling Mad localization. We further confirmed that moleskin (msk 5 ) interacts with Dg and Dys ( Table 1).
The next largest class of modifiers (9/27 EMS induced modifiers; Table S1) belonged to the Suppressor+ (Su+) class with extra wing vein material ( Figure 3) and showed a more global effect on the wing. Four of these modifier mutants, Mod59, Mod111, Mod130 and Mod140, had phenotypes in the absence of   Figure S3; 2, summary for the data in Figure 4; +, interact; -, does not interact. doi:10.1371/journal.pone.0002418.t001   the dystrophin mutant. We mapped them to cytological location 92A and showed that they are tightly linked to the Delta gene (Dl, Table 1). Subsequent crosses to different Dl alleles yielded lethal phenotypes, suggesting that these mutants are alleles of Dl. Two of the modifiers Mod130 and Mod140 suppressed the posterior crossvein phenotype of Dys N-RNAi , i.e. formed a complete crossvein from longitudinal vein L4 to longitudinal vein L5, in addition to generating extra wing vein material. We also found modifiers that belonged to the Enhancer as well as the Suppressor classes (Text S1).
Two other members of the class of mutants with extra wing vein material (Table S1 and Figure S2B), Mod4 and Mod29 were mapped to specific locations on the 2nd chromosome. Mod4 was mapped by following its lethality phenotype in crosses with deficiency lines. It was lethal in a cross with the Df(2R)nap9/ Dp(2;2)BG, In(2LR)Gla line and was not lethal in a cross with the Df(2R)ST1, Adh[n5] pr [1] cn[*]/CyO line. The available lethal mutants from cytological region 42A1-42B3 were crossed to Mod4 and using a P-element insertion line, PBac{PB}CG7845c00845/ CyO we determined that the mutation was in gene CG7845. This gene codes for a WD40 containing protein whose function is not known ( Table 1). The underlying common function of all WDrepeat proteins is coordinating multi-protein complex assemblies, where the repeating units serve as a rigid scaffold for protein interactions. Such a scaffolding protein may be utilized in the formation or stabilization of the DGC.
Following the wing vein phenotype seen in homozygous flies as well as its semi-lethality, Mod29 was mapped to cytological region 27D1-27D4. This region was independently identified to contain a Dys interactors through the Deficiency screen ( Figure 5B). We further fine-mapped the region by showing that two smaller deficiency lines, Df(2L)ade3/CyO,P{ftz/lacB}E3 and Df(2L)Ex-el7029,P+PBac{XP5.WH5}Exel7029/CyO phenocopy the semilethality/wing vein phenotypes seen with Mod29. Ultimately, Mod29 was determined to be a poly-EGF gene mutant. Other genes in the region were eliminated by complementation analysis. In addition to the wing vein phenotype, Mod29 also shows defects in muscles, in oogenesis and in the brain (photoreceptor axon termination; Figure S5). Additionally, these mutant animals have a very short lifespan. Since Poly-EGF is predicted to bind the Notch receptor further functional analysis will be very interesting.
As discussed above, the Dys E6 mutant does not show a wing vein phenotype as a heterozygote (Figure 2A). But as a homozygote it displays a disrupted posterior crossvein similar to Dys N-RNAi ( Figure 2B and 2C). We reasoned that this heterozygote background was ''sensitized'' and therefore excellent for identifying enhancers. We screened 6000 chromosomes and found four modifiers for Dys E6 /+. Three of those modifiers showed some extra wing vein material in the posterior crossvein ( Figure S2B). Two of the modifiers, on the third chromosome ModE21 and ModE26 showed a branched crossvein phenotype in the absence of Dys E6 though this dominant phenotype was not completely penetrant. The third of this group, ModE11 on the second chromosome, showed no wing vein phenotype in the absence of Dys E6 , however as a homozygotes ModE11 flies displayed a globally disrupted wing vein characterized by an abundance of extra vein material (data not shown). Modifier E10 was the only modifier that showed an enhanced phenotype ( Figure S3A). ModE10/Dys E6 flies phenocopied the Dys E6 homozygous phenotype which is a reduction in wing vein material ( Figure 2B). Moreover, ModE10/Df(3R)Exel6184 flies also gave a similar wing vein phenotype to Dys E6 /Df(3R)Exel6184 flies strongly arguing that ModE10 is a dystrophin allele. Finding an additional allele of Dys confirms that the screening approach was successful and analysis of the EMS mutant will allow us to further characterize Dystrophin protein function. Furthermore, the newly identified dystrophin allele ModE10 interacted strongly with the Dg RNAi mutant in the wing; instead of 27% penetrance, the phenotype was 96% penetrant in heterozygous Dys[Mod-E10] background ( Figure 4). This further confirms that molecules acting closely with Dys and Dg can be found in this assay.
mbl Interacts with Dys in Muscles These modifier screens led to the identification of genes that belong to different functional groups. Importantly, genes with known muscle function were identified in this screen, including dystrophin itself. In addition, the screen also identified muscle genes previously shown to interact with the Dg-Dys complex, Cam, mbl and nAcRa-30D [22,23,24]. muscleblind encodes a RNA binding protein that has shown to function in splicing [25]. As part of a secondary screen in muscle tissue, we tested whether mbl, a strong interactor with Dys in wing vein, also interacts with Dys in the indirect flight muscles.
Animals that have lost one copy of mbl appear to have normal muscle structure ( Figure 6A), while act-Gal4:UAS-Dys N-RNAi mutants show moderate muscle degeneration (10 days old: 10%, n = 68; 18 days: 25%, n = 159; Figure 6B and 6D). However, the extreme muscle degeneration phenotype was not observed in 10 days old and at a low frequency in 18 day old Dys N-RNAi mutants ( Figure 6B and 6D). In contrast, when mbl was reduced by one copy in a Dys N-RNAi mutant background, increase in muscles defects was observed. Furthermore, around 50% of the phenotypes observed were classified ''extreme'' both in 10 and 18 day timepoints ( Figure 6C and 6D). These data suggest that mbl and Dys interact in the muscles and reduction of muscleblind level enhances the abnormal muscle phenotype in Dys N-RNAi mutants. These results show that the screen successfully identified genes that interact with dystrophin to establish normal muscle function.

Interactors Show Germline Phenotypes Similar to Dys and Dg Mutants
The Dys and Dg genes are required in the germ line for the establishment of oocyte polarity [6,8,9]. The oogenesis defects of Dys and Dg serve as an excellent test to identify whether the new modifiers might interact with Dg-Dys complex in other tissues as well as in wing development. The ovariole contains a progression of egg chambers at different stages of development ( Figure 7A). An early oocyte polarity marker, Orb, is localized to the anterior side in early stages and then migrates to the posterior side of oocyte by stage 3 of oogenesis. Between stages 3 and 6, Orb is clearly localized to the posterior of the oocyte, making it an excellent marker to analyze the polarity of the oocyte ( Figure 7A-B). Dg lossof-function germ line clones and homozygous point-mutants are arrested at early stages of oogenesis ( [9]; Figure 7B) and show mislocalization of the Orb marker that is usually missing or diffused ( [9]; Figure 7D). The oocyte polarity phenotype was also observed for viable Dg and Dys alleles ( Figure 7B, 7D and 7E; 74%, n = 50 for Dg O86 /Dg O55 , 82%, n = 34 for Dg O43 /Dg O55 and 68%, n = 48 for Dg O86 /Dg O43 ; the ovaries were dissected from 12 days old mutants).
Interestingly, many of the identified modifiers are genes previously shown to be required in germ line development. For example a scaffolding protein, Rack1 that contains multiple WDdomains and interacts with atypical protein kinase C (aPKC) has recently been shown to function during Drosophila oogenesis [26]. We now analyzed potential Rack1 function in germ line for posterior Orb localization. Importantly, this analysis revealed that Rack1 is required in the early oocyte polarity: lack of Rack1 in early oocytes resulted in Orb mislocalization in the majority of the eggchambers (76%, n = 13, Figure 8H). Consistent with these findings aPKC is required for early oocyte development in Drosophila [27]. Similarly, another Dys and Dg interacting protein, Deadlock has previously shown to be required for Orb localization [28]. It is, therefore, likely that other identified Dys and Dg interactors are required in early oocyte polarity, possibly in a process closely related to Dg-Dys function in oogenesis.
Utilizing the FRT sites in the P-lethal lines we made germ line clones to determine how many of the selected modifiers had a role in the establishment of early oocyte polarity. We analyzed germ line clones of 14 loss-of-function mutants and found 8 genes that showed a similar phenotype to that seen in Dg loss-of-function clones. One of these genes, POSH, is an SH3 adapter protein.
Loss-of-function clones of POSH ( Figure 8A, white arrow) were arrested prior to stage 3-4 of oogenesis and appeared to lack Orb entirely. Orb mislocalization was also found in Sema-2a, kek1, kis, Lis-1, chif and Khc. Sema-2a loss-of-function clones showed clear developmental arrest prior to stage 4-6 with diffused Orb staining ( Figure 8B). kek1, a negative regulator of the EGFR pathway, also showed developmental arrest in loss-of-function clones ( Figure 8C). Orb in kek1 clones surround the oocyte in an irregular pattern. kis loss-of-function germline clones have shown developmental arrest and diffused Orb staining ( Figure 8D). Similarly, the majority of Lis-1, chif and Khc germline clones showed abnormal Orb localization ( Figure 8E-8G). Lis-1 and Khc have previously shown to be involved in oocyte microtubular motor activities and to interact in this process [29,30].
These data indicate that many of the genes that showed interactions with Dys and/or Dg in the wing vein assay, also showed early oocyte polarity phenotypes similar to Dys and Dg and potentially play a role in the establishment of early oocyte polarity.

Discussion
In mammalian systems the DGC is composed of Dystrophin, Dystroglycan, the Sarcoglycan complex (a, b, c, d, e, and f), Sarcospan, the Syntrophins (a and b), the Dystrobrevins (a and b) and Caveolin-3 [2]. Presently, the regulation of Dg-Dys complex and its involvement in signaling are poorly understood. In this work, we have addressed these unknowns by using dominant modifier screens to find genes that may shed light on both of these processes. Our screens have revealed groups of modifiers that are components of canonical signaling pathways (TGF-b, EGFR, Wnt and Notch) as well as components involved in cell/neuronal migration, cytoskeletal rearrangements and most importantly muscle development.

New Components of the Dg-Dys Pathway?
The screens described in this paper revealed some expected interactors, Dys, Cam and Khc. Calmodulin, a calcium binding protein required for muscle and neuronal functions has previously been shown to interact with mammalian the Dg-Dys complex. However, whether reduction of Cam activities suppresses or enhances the muscular dystrophy phenotype is not totally clear. Targeted inhibition of Cam signaling exacerbates the dystrophic phenotype in mdx mouse muscle while genetic disruption of Calcineurin improves skeletal muscle pathology and cardiac disease in ä-sarcoglycan null mice [31,32]. Since reduction of Cam showed suppression of the phenotypes caused by reduction of the long forms of dystrophin in the Drosophila wing, it will be interesting to analyze whether reduction of Cam will suppress the Drosophila Dys muscle phenotype as well. Khc involvement in Dg-Dys complex is also expected since work in mammalian system has shown that Khc can bind Dystrobrevin, a component of Dg-Dys complex [33]. It will be interesting to test in the future whether Drosophila Dystrobrevin can similarly bind Khc and what the functional significance of this interaction is in muscles and neurons. We have already shown that in oocyte development Khc is required for the same early as Dys and Dg (Figure 7). It is, therefore, interesting to test the potential requirement of dystrobrevin in this process and to further dissect the Khc function in this complex during early polarity formation.
An unexpected new interactor was identified in our screens, the homeodomain interacting protein kinase, HIPK. In mammalian systems HIPK is involved in the cell death pathway by phosphorylating p53 [34]. Recently Drosophila HIPK has shown to be involved in a communal form of cell death, sudden, coordinated death among a community of cells without final engulfment step [35]. It remains to be seen whether this HIPK communal death pathway will utilize p53 phosphorylation. However, it is tempting to speculate that the cell death observed in muscular dystrophies use the newly described HIPK dependent communal death pathway. HIPK is shown to interact with a WD40-protein in mammalian system [35]. Since three WD40 proteins were identified in our screens, it will be interesting to test whether any of these interact with Drosophila HIPK.
Another interactor that might shed light in the pathways utilized by the Dg-Dys complex is an SH3-domain adapter-protein, POSH. Structure-function analysis of Dg protein has revealed that a potential SH3-domain binding site in Dg C-terminus is essential for Dg function [8]. However, the critical SH3-domain protein in this complex is still at large. The present screen revealed that POSH can interact with the Dg-Dys complex in the wing vein assay. It will now be interesting to determine whether POSH is the missing critical SH3-domain protein that interacts with Dg-Dys complex in Drosophila.

The Neuronal/Cell Migration Pathways Interact with the DGC
There are only a few examples of signaling pathways that have been shown to transmit information from outside the cell that results in cytoskeletal rearrangements inside the cell. Slit-Robo, Netrin-Frazzled and Semaphorin-Plexin pathways are examples of such activity. Dg-Dys complex appears also regulate the cytoskeleton based on extracellular information. Interestingly, the interaction screens described in this paper show that these aforementioned pathways are much more intricately connected than previously thought. The Robo and Netrin Receptor (DCC) pathways have previously been shown to interact [36], now we report that Dg-Dys complex interact with these pathways as well.
The interactions that we see in wing development involving the Drosophila DGC and the genes that affect neuronal guidance (sli, robo, fra, sema-2a, sema-1a, Sdc) might be explained by their possible role in hemocyte (insect blood cell) migration. Analysis done in Drosophila shows that known axon guidance genes (sli, robo) are also implicated in hemocyte migration during development of the central nervous system [37]. Similar findings have been reported in mammals, where blood vessel migration is linked to the same molecular processes as axon guidance [38]. Both sli and robo have been implicated in the vascularization system in vertebrates [39]. A recent study demonstrated that proper hemocyte localization is dependent upon Dys and Dg function in pupa wings. Mutations in these genes result in hemocyte migration defects during development of the posterior crossvein [14]. Hence, we speculate that the neuronal guidance genes that we have found may interact with the DGC in wing veins by having a role in the migration process.
Similar to sli and robo [40], the Dys and Dg mutants also affect photoreceptor axon pathfinding in Drosophila larvae [6]. It is therefore possible that this group of modifiers will interact with the DGC in axon pathfinding and other processes. Supportive of that notion is the fact that mammalian Syndecan-3 and Syndecan-4 are essential for skeletal muscle development and regeneration [41]. In addition slit-Dg interaction has previously been observed in cardiac cell alignment [42]. Sequence analysis of slit reveals that it possesses a laminin G-like domain at its C-terminus. Dystroglycan's extracellular domain has laminin G domain binding sites and has been shown to bind 2 of the five laminin G domains in laminin. It is therefore possible that slit, through its laminin G-like domain, binds to Dystroglycan and that Dystroglycan is a slit receptor. It will be informative to reveal the mechanisms and nature of these interactions.

Oocyte Phenotypes
The establishment and formation of oocyte polarity during development is a complex multistep process (reviewed in [43]). In the anterior part of the germarium each stem cell undergoes asymmetric cell division to give rise to another stem cell and a cystoblast. The cystoblasts divide four times with incomplete cytokinesis to form a 16 cell cyst. The oocyte fate is determined when the cyst reaches the end of the germarium. At this point, BicD protein, Orb protein, the microtubule organizing center (MTOC) and the centrioles move from the anterior to the posterior of the oocyte ( Figure 7A). These events mark the first sign of polarity in the oocyte. Subsequent Gurken signaling induces posterior follicle cells to signal back to the oocyte which repolarizes the microtubule cytoskeleton. This signal appears to require an intact extracellular matrix since Laminin A mutants do not undergo repolarization [44]. The outcome of the repolarization results in the disassembly of the MTOC at the posterior, nucleation of microtubules anteriorly and subsequent migration of the oocyte nucleus to an antero-lateral position.
Germ line clones that lack Dg show developmental arrest and mislocalization of the oocyte polarity marker Orb which is usually diffused or absent in the oocyte [9]. This phenotype might be due to Dg affecting the localization of the MTOC. But how exactly Dg is involved in this process is not clear. One possible explanation is that Dg is required for extracellular matrix (ECM) integrity since it is known to bind Laminin. Such a structural conduit may be necessary for proper signaling from the posterior follicle cells to the oocyte. This is supported by the fact that Dg loss-of-function mutants show defects in Actin accumulation. Another possibility is that Dg may be involved in mircrotubule organization. Since the regulation of actin-and microtubule-cytoskeleton are connected, these two models are not mutually exclusive.
Interestingly, in our screens we found several genes that showed phenotypes in oocyte development. One of these genes is kek1, a transmembrane protein of the Drosophila Kekkon family that has been reported to be a negative regulator of the EGFR receptor [45]. It has been previously shown that EGFR signaling regulates the expression pattern of Dystroglycan to establish anteriorposterior polarity of oocyte [10]. Further study is required to determine if kek1, as an EGFR regulator controls Dg expression in the germ line.
Another interesting gene found in our screens is POSH (Plenty of SH3 domains), a Drosophila homologue of human SH3MD2 protein. Interestingly POSH is a multidomain scaffold protein that can interact with Rho related GTPase -Rac1 and promotes the activation of the JNK pathway [46] POSH has also shown to regulate POSH-MLK-MKK-JNK complex [47]. A defect in this complex can affect brain function. Furthermore, POSH and JNKmediated cell death pathway is thought to play an important role in Parkinson's disease [48]. With many SH3 domains, POSH has the potential to bind Dg that has a predicted SH3-domain binding site and has been shown to be necessary for the establishment of oocyte polarity [8].
In addition, we have found interactions with Khc, Lis-1 and Dmn, three genes known to be part of the Dynein-Dynactin complex which in addition to Kinesin microtubule motor activity have been shown to be necessary for establishment of intracellular polarity within the Drosophila oocyte [49,50]. In mid-oogenesis dynein, dynactin and kinesin are thought to act cooperatively in cargo transport [49,51,52]. Since these genes interact with Dys (Table 1) and show similar phenotypes in Orb localization (Figure 7), it will be interesting to dissect their potential functional interactions with Dys in early oocyte development. Furthermore, since mammalian Dystrobrevin physically interacts with Khc, it is plausible, that the Dynein-, Dynactin-, Kinesin-complex will utilize localization cues set-up by Dg-Dys Complex.
In addition to the interactions with microtubular motorcomplexes, we also found interactions with a Drosophila Formin homologue, FHOS. Mammalian FHOS directly binds to F-actin and promotes actin fiber formation [53]. Recently Drosophila actin nucleators, Capu and Spire have shown to assemble a cytoplasmic actin mesh that maintains microtubular organization in the middle of oogenesis [54,55,56,57,58]. Therefore, it will be important to determine whether the actin nucleator, FHOS is also involved in actin nucleation that regulates microtubular activity in early oocyte development. Further study of these cytoskeletal genes will allow us to gain a more detailed understanding of how Dg and Dys function to ensure proper oocyte polarity during oogenesis.
Similar to microtubule and actin interplay in the regulation of oocyte polarity, the dynamic actin-microtubule interactions regulate growth cone steering at the growing axons [59]. It is therefore possible that similar mode of function for Dg-Dys interactions with these cytoskeletal modules is used in various cell types. Furthermore the axon pathfinding and oocyte polarity formation processes are similar in that they are controlled by extracellular information which is transmitted to the cell resulting in cytoskeletal rearrangement.

Planar Cell Polarity Genes
At the basal side of follicle epithelium, actin filaments exhibit a planar cell polarity that is perpendicular to the long axis, the AP axis, of the egg chamber. In Dg follicle cell clones the basal actin array is disrupted non-cell-autonomously [9]. Integrins and the receptor tyrosine phosphatase Lar are also involved in basal actin orientation [60]. It is unclear whether Dg and the other genes involved in basal actin polarity act together with the canonical planar cell polarity pathway or function independently of this pathway. Interestingly, we now report strong interactions between the DGC and grainy head (grh) a transcription factor which is required for several different processes during the differentiation including the function of the frizzled dependent tissue polarity pathway, epidermal hair morphogenesis and wing vein specification [61]. In the absence of grh function the Fz, Dsh and Vang proteins fail to accumulate apically and the levels of Stan (or Flamingo) protein are dramatically decreased. The interactions seen with stan (Fla) and wg in wing veins supports the hypothesis that Dg might act together with the frizzled-dependent tissue polarity pathway in coordinating the polarity of cells in epithelial sheets.

Conclusions
By screening for alterations of a dominant wing vein phenotype we have found modifiers of the DGC that are involved in cytoskeletal organization. Initial characterization of some of these genes revealed that they have phenotypes also in other tissues, in which the DGC is known to function. These tissue/cell types include the oocyte, the brain and the indirect flight muscles. This argues strongly that the identified interactors may be involved globally in DGC function. Further study is required to determine mechanistically how these modifiers work in the context of the Dg-Dys complex. However a common theme, already arising is that the identified interactors appear to regulate cytoskeletal rearrangement. Mechanistic understanding of how the new interactors might regulate Dg-Dys communication with cytoskeleton of muscle cells may serve as a basis for the development of novel therapeutic approaches that might improve the quality of life of individuals afflicted with muscular dystrophy.
The following Dys and Dg mutants were used to screen for modifiers of the wing vein phenotype: Dys E6 , an hypomorphic allele of dystrophin, act-Gal4:UAS-Dys N-RNAi /CyO, tub-Gal4:UAS-Dys N2-RNAi /TM3, tub-Gal4:UAS-Dys C-RNAi /TM3 which are three Dys RNAi mutants recombined onto chromosomes with actin and tubulin Gal4 drivers, respectively; and tub-Gal4:UAS-Dg RNAi /TM3, a Dg RNAi mutant recombined onto a chromosome with the tubulin Gal4 driver. Dys N-RNAi is a transgene on the second chromosome that encodes an inverted repeat that forms a double stranded RNA hairpin complementary to the corresponding Nterminus of the protein. It reduces the protein levels of all the known long isoforms. Dys N2-RNAi is a transgene on the third chromosome that encodes a different inverted repeat than Dys N-RNAi but still reduces the protein levels of the known long isoforms. Dys C-RNAi is a transgene on the third chromosome that encodes sequence complementary to the corresponding C-terminal region of dystrophin and reduces the protein levels of all isoforms. All Dys RNAi mutants exhibited 100% penetrance of the wing vein phenotype. The Dg RNAi mutant exhibited ,30% penetrance of the wing vein phenotype.
P-element screen. The FRT P-element insertion lethal lines used in this study were obtained from the Kyoto Stock Center (Japan). From this collection, 800 lines were screened. Two-to-four day-old males carrying P-lethal insertions over balancers (or virgin females if the P-element was on the X chromosome) were mated to Dys E6 /TM3; act-Gal4:UAS-Dys N-RNAi /CyO; tub-Gal4:UAS-Dys C-RNAi / TM3 and tub-Gal4:UAS-Dg RNAi /TM3 virgin females (or males). Non-balancer F 1 progeny were screened for dominant modification of the wing vein phenotype. Modifiers were divided into phenotypic classes based on the alterations of their wing veins as compared to wing veins of act-Gal4:UAS-Dys N-RNAi /(+ or CyO), tub-Gal4:UAS-Dys C-RNAi /(+ or TM3) and tub-Gal4:UAS-Dg RNAi /(+ or TM3) sibling flies. Suppressors were identified when wing veins of F 1 flies phenocopied the wing veins seen in wild type flies. Since Dys E6 /+ had wild-type wing veins, non-balancer F 1 progeny from the Dys E6 /TM3 and P-lethal insertion cross were analyzed for the presence or absence of a mutant posterior crossvein which is altered in Dys E6 homozygotes. All the genes found in this screen were crossed to w[1118] to verify the absence of a dominant wing vein phenotype. None modifiers showed a dominant posterior wing vein phenotype. Since P-element insertion lethal lines were on chromosomes with FRT sites, we made mosaics and determined if selected modifiers possessed an oocyte polarity phenotype (see below).
EMS screen. For the EMS mutagenesis 2-4 day old w 1118 males were place in yeasted bottles overnight. Twenty males were placed in each vial and starved for 6-8 hours. The flies were then provided access to a solution of 22-25 mM EMS (ethyl methansulfonate; Sigma) in a 5% sucrose solution for 18 hrs. The flies were placed in new vials and allowed to recover for 1-3 hrs. Each vial of flies was transferred to yeasted bottles with 40 virgin females of the desired genotype per bottle. The females were allowed to lay eggs for two days then transferred to new bottles daily for two more days. Mutagenized w 1118 males were crossed to act-Gal4:UAS-Dys N-RNAi /CyO or tub-Gal4:UAS-Dys N2-RNAi /TM3 females. Males of these crosses were scored for a modification of the wing vein phenotype, either a novel alteration or suppression. These were picked and backcrossed to the parental Dystrophin mutant females to determine whether the modification was on the second or third chromosomes. Once this was done these modifiers were balanced either with CyO or TM6B and maintained as a stock. The subset of modifiers that produced phenotypes in the absence of the Dys mutants were crossed to meiotic mapping stocks and subsequently mapped using those phenotypes. The remainders were crossed to meiotic mapping stocks with a dystrophin mutant in the background and mapped using the modification phenotypes. In addition to the two RNAi dystrophin mutants mentioned above, enhancers of the Dys E6 mutant were also screened for.
Obtained modifiers were mapped meiotically using the lines from Bloomington Stock Center al [1]

Loss of Function Mosaic Analysis
To obtain germline clones of the modifiers found in the FRT Pelement lethal screen, they were crossed to corresponding chromosomal FRT stocks. Third instar larvae and early pupae from this cross were heat-shocked at 37uC for 2 hours. After eclosion, they were placed in vials with fresh yeast paste for 4-5 days before dissection.

Dissections and Immunohistochemistry
Drosophila ovaries and wing imaginal discs were dissected rapidly in phosphate buffered saline (PBS) and fixed in 4% paraformaldehyde (PFA) for 10 minutes at room temperature. Adult's brains were dissected in PBS and fixed in PFA for 30 minutes. All antibody stainings were performed as described previously [6], except with the Dystrophin antibody with which a protocol provided by L. Fradkin was used [62]. Ovaries, brains and wing imaginal discs were mounted in 70% glycerol in PBS for analysis using a Leica TSC SPE confocal microscope. The following antibodies were used: rabbit anti-Dg (1:3000; [9]), anti-Dys CO 2 H (1:3000; [62]), mouse anti-Orb (1:20; Developmental Studies Hybridoma Bank), mouse anti-24B10 (1:20; Developmental Studies Hybridoma Bank), Alexa 488, 568, or 633 goat antimouse, Alexa 488, 568 goat anti rabbit. To mount adult wings, flies were pre-incubated in 80% ethanol and 20% glycerol solution, then dissected and mounted in 70% glycerol before analysis using a Leica light microscope.

Muscle Analysis
Histological sections of muscle were prepared from waxembedded material as described previously [6], stained with hematoxyline and eosin (H&E staining) and analyzed using light microscopy.

Table S1
Found at: doi:10.1371/journal.pone.0002418.s002 (0.14 MB DOC) Anterior to the left. The germarium is at the anterior most tip of the ovariole with developing egg chambers progressing to the right (posteriorly). The egg chambers undergo developmental arrest in later stages. Actin-Green, Adducin-Red, DAPI-purple. (D) 24B10 antibody staining of the adult wild type brain. Arrows indicate photoreceptor axon termination sites where the R8 photoreceptor axon (left arrow) terminates before the R7 axon (right arrow). (E) Mod29/Df adult brain. Termination of the R8 and R7 axons are indicated by the red arrows. White arrows indicate non termination of two axons that protrude deeper into the brain. (F) Dg323/Dg323 clone in the adult brain. Red arrows indicate where the R8 and R7 axons should terminate. Upper white arrows indicate a general disruption of axon termination in the R8/R7 termination region. Lower right arrow indicates a non terminating axon that proceeds deeper into the brain. Found at: doi:10.1371/journal.pone.0002418.s007 (3.58 MB TIF)