Evaluating a potential interaction between deltex and git in Drosophila: genetic interaction, gene overexpression and cell biology assays. The data described in chapter 3 presented evidence that endogenous dx is required for proper myogenesis in the fly embryo. The data also showed that overexpression of dx or overexpression of the dominant-negative form dx ΔPRM in the embryonic mesoderm affect various aspects of muscle development in the embryo and lead to pupal lethality. In this chapter, evidence for a potential interaction between dx and another fly gene, git, is presented and assessment of the interaction between these two genes was based on the type of embryonic muscle phenotypes, pupal lethality and adult phenotypes produced in various genetic combinations. An interaction between Dx and GIT was first reported in a yeast two-hybrid system designed to identify a genome-wide protein interaction map for Drosophila (Giot et al., 2003). GIT belongs to a family of scaffolding proteins that were first identified in mammals (Premont et al., 1998). GIT proteins have a complex domain structure and appear to have important functions in the control of focal adhesion disassembly, cytoskeletal dynamics, cell migration and membrane trafficking between the plasma membrane and recycling endosomes (Manabe et al., 2002 ). GIT comprises an aminoterminal zinc finger-like motif, a N-terminal Arf-GAP domain for converting active Arf- 68
GTP to inactive Arf-GDP, ankyrin repeats for protein-protein interaction, a Spa2- homology domain which binds FAK and PIX, a coiled-coil domain and a conserved carboxyl-terminal region that interacts with paxillin and links GIT to the focal adhesion complex (Hoefen et al., 2006). In Drosophila, there is a single GIT homologue, dgit (FLYBASE:FBgn0033539). dgit mutants were isolated in our laboratory and it was shown that dgit is expressed in embryonic muscles and localizes to muscle attachment sites (Bahri et al., unpublished). dgit mutants are semi-lethal and result in muscle defects in the embryo as well as adult phenotypes such as ectopic bristles and defective wings (Bahri et al., unpublished). In this chapter, I present evidence that dx and dgit interact genetically in a positive manner to affect adult phenotypes such as wing posture and bristle number, suggesting that Dx and GIT may act in the same genetic pathway to affect these adult structures. In addition, the data from overexpression experiments showed that Dx and GIT may also interact in a negative manner where GIT counteracts Dxoverexpression effects in embryonic muscles and fly development. 4.1. Genetic interaction between dx and dgit mutant alleles 152 21C 4.1A. Genetic interaction between dx and dgit Genetic interaction between two genes is generally characterized by the phenotypic analysis of animals lacking either, one copy of each of the two genes, lacking two copies of one gene and one copy of the other gene, or lacking two copies of both genes. To test if dx genetically interacts with git, virgin female flies homozygous for dx 152 null allele were crossed to male flies homozygous for dgit 21C and their male and female progenies were collected and phenotypically scored (Fig. 4.1A.1). All male progeny from 69
this cross lack both copies of dx and only one copy of git (dx 152 /y git 21C /) while all female progeny lacks only one copy of dx and one copy of git (dx 152 / git 21C /). Should there be an interaction between dx and git, phenotypes in the progenies would be expected. Indeed, a new wing phenotype (that was not seen in the parents) was observed in 80% of the male progeny (Fig. 4.1A.2 Table 4.1A.1), suggesting that dx and git genetically interact the male wings were held-out at 45 degrees from their body axis while control males which lacked both copies of dx but still had both copies of git intact did not show defective wing posture, indicating that the wing posture phenotype was due to the removal of both copies of dx and one copy of git. In contrast to the male progeny, the wing posture of the female progeny (which lack only one copy of dx and one copy of git) was normal, again suggesting that both copies of dx and one copy of git have to be removed in order for the held-out wing phenotype to be expressed. Mutations causing held-out wings have previously been isolated and characterized as affecting muscles that function in wing retraction during rest-phase of the adult fly (Baehrecke, 1997 Soanes and Bell, 1999). Similarly, the held-out wing phenotypes in dx 152 /y git 21C / males may be caused by muscle defects. The data from this experiment showed that dx and git genetically interact with each other and suggest that Dx and GIT may be working together (directly or indirectly) in a genetic pathway that probably affect adult muscles and consequently wing posture. Alternatively, the complete removal of Dx 70
Genetic cross between deltex and git mutants: dx 152 dx 152 X y git 21c git 21c dx 152 y git 21c & dx 152 git 21c 80% males having held-out wings Figure 4.1A.1. Genetic interaction between dx 152 and git 21C. The diagram shows a genetic interaction cross between homozygous dx 152 mutant females and homozygous git 21c mutant males. 80% of the male progeny have held-out wing phenotypes, but the female progeny was largely normal. Held-out wing Vein Genotypes phenotypes Defects dx 152 /y dgit 21c / / (interaction males) 80% (n=133) 100% (n=133) dx 152 /y/ / (control1 males) 0% (n=100) 100% (n=100) /y dgit 21c / / (control2 males) 0% (n=100) 0% (n=100) dx 152 / dgit 21c / / (interaction females) 0% (n=126) 0% (n=126) dx 152 / / / (control1 females) 0% (n=100) 0% (n=100) / dgit 21c / / (control2 females) 0% (n=100) 0% (n=100) n = number of adult flies Table 4.1A.1. Phenotypes of dx 152 /git 21C transheterozygous flies. 80% of interaction males lacking both copies of dx and one copy of git exhibit a held-out wing phenotype but controls or interaction females do not show this phenotype the 100% delta vein defect associated with dx 152 males was not modified by the removal of one copy of git. Percentages of wing phenotypes were calculated as follows: number of flies with wing defects / total number of flies x100. 71
A B Figure 4.1A.2. Held-out wing phenotype in dx 152 /git 21C interaction. Panel A shows males of the genotype dx 152 /y dgit 21c / displaying held-out wing phenotypes panel B shows wt control flies with normal wing postures. 72
may weaken the male flies and make them sensitive to additional mutations (like in the case of removal of one copy of git) and hence additional reduction of git by half in dx 152 mutant males now produces new phenotypes that were not observed in dx 152 mutant males alone. 1 21C 4.1B. Genetic interaction between dx and dgit Instead of dx 152, the dx 1 allele was used in another genetic interaction cross with git 21C (Fig. 4.1B.1). 47% of female progeny (dx 1 / git 21C /) lacking one copy of dx and one copy of git and 15% of male progeny (dx 1 /y git 21C /) obtained from this, displayed ectopic bristles on their scutella (posterior sections of adult thorax Fig. 4.1B.2) it is not clear why transheterozygous females showed higher frequency of bristle phenotypes than the males, probably due to the nature of the dx 1 allele the molecular defect in dx 1 has not been yet determined. wt flies normally have 4 mechanosensory bristles on their scutella, while the transheterozygous flies often showed six or more mechanosensory bristles (Fig. 4.1B.2). The ectopic bristles in these animals were associated with ectopic sockets, suggesting that they are probably complete organs and not a result of cell fate transformation in the same organ (for example a socket cell taking the fate of a hair cell) a mechanosensory organ is normally made of an exterior hair cell, an exterior socket cell, and interior neuron and support cells (Walker et al., 2000). Interestingly, homozygous git 21C flies have been shown to display a similar ectopic mechanosensory bristle phenotype (Bahri et al., unpublished). Since the bristle phenotype was observed in animals lacking one copy of dx and one copy of git in the interaction cross, this may 73
indicate that Dx and GIT act in the same genetic pathway to affect bristle number. Control flies lacking only one copy of dx or control flies lacking Genetic cross between deltex and git mutants: dx 1 dx 1 X y git 21c git 21c git 21c git 21c dx 1 dx & 1 y 15% males and 47% females having ectopic bristles Figure 4.1B.1. Genetic interaction between dx 1 and git 21C. The diagram shows a genetic interaction cross between homozygous dx 1 mutant females and homozygous git 21c mutant males. 15% (n=20) of the male progeny and 47% (n=36) of the female progeny display ectopic bristle phenotypes on their scutella. A B SC SC 74
Figure 4.1B.2. Ectopic bristle phenotype in dx 1 / dgit 21c interaction experiment. A shows a female of the genotype dx 1 / dgit 21c / with ectopic bristle phenotype on its scutellum (arrows) B shows a rescued female of the genotype dx 1 / dgit 21c /su(dx) with normal bristle phenotype on its scutellum (arrow) SC: scutellum. only one copy of git did not show the ectopic bristle phenotype, suggesting that this phenotype is due to the simultaneous removal of one copy of dx and one copy of git in the same animal. To determine whether this interaction is specific to dx, I made use of the available fly line dx 1 / dx 1 Su(dx)/Su(dx) Su(dx) is known to suppress the dx phenotypes (Matsuno et al., 2002). This line was used to repeat the genetic interaction cross with git 21C females and the female progeny of the genotype dx 1 / Su(dx)/git 21C were inspected. These females did not show ectopic bristles on their scutella (Fig. 4.1B.2B) indicating that Su(dx) is suppressing the dx 1 / git 21C interaction bristle phenotype. The data also suggests that the ectopic bristle phenotype is due to the interaction between dx 1 and git 21C. It is interesting to note that the adult phenotypes observed in the interaction experiment between git and dx 1 were different from those observed in the interaction experiment between git and dx 152, suggesting that dx 1 and dx 152 alleles behave differently for example dx 1 heterozygous females that lack one copy of git display ectopic bristles (47%, n = 36) whereas dx 152 heterozygous females that lack one copy of git do show this phenotype but at a lower frequency ( 4.5%, n = 62) similarly, dx 1 hemizygous males that lack one copy of git display ectopic bristles (15%, n =20 ) whereas dx 152 hemizygous males that lack one copy of git rarely show this phenotype (1%, n = 85). In addition, dx 152 hemizygous males that lack one copy of git display wing posture defect whereas dx 1 hemizygous males that lack one copy of git do not show this phenotype. This qualitative difference between dx 152 and dx 1 alleles is also corroborated with the immunostaining 75
results described in the previous chapter the immunostaining data showed that dx 152 is null (as expected from published reports) whereas dx 1 still encodes a Dx form that, unlike Dx, can localize primarily to muscle attachment sites (see chapter 3 Section 3.3). The molecular lesion in dx 1 has not yet been determined and it will be interesting to map it in the future. 4.2. Genetic interaction between dx and git in overexpression assays 4.2A. Pupal lethality associated with Dx-overexression under mesodermal driver was suppressed by GIT overexpression The data described in chapter 3 showed that mesodermal overexpression of Dx causes 100% pupal lethality. In order to test for interaction between Dx and GIT, Dx was overexpressed together with GIT under 24bgal4 in the same animal and the extent of pupal lethality was scored (Fig. 4.2A.1 Table 4.2A.1)). In this experiment, 22% of the animals that are simultaneously overexpressing one copy of dx and one copy of git hatched from their pupal cases and looked normal (Fig. 4.2A.1 Table 4.2A.1), while few of them hatched half-way and remained stuck in their pupal cases (3%, Fig. 4.2A.2C) the rest of the animals from this experiment did not hatch and died as pupaes. The data from this experiment suggested that overexpression of GIT may antagonize the toxic effect of Dx overexpression and partially rescue the pupal lethality associated with Dx overexpression alone. The half-way hatching phenotype observed in the rescue experiment may be indicative of muscle defects, as it is generally thought that adults that 76
display such hatching phenotypes are too weak to be able to push themselves out of the pupal case. Overexpression of deltex in combination with git using 24bgal4 driver: 24bgal4-uas-gitflgfp 24bgal4-uas-gitflgfp X y uas- dx-gst uas- dx-gst & uas- dx-gst 24bgal4-uas-gitflgfp uas- dx-gst 24bgal4-uas-gitflgfp y 22% of pupae hatch into adults. Figure 4.2A.1: Suppression of pupal lethality associated with Dx overexpression by GIT overexpression. The diagram shows a genetic interaction cross between homozygous uas-dx-gst males and homozygous 24bgal4-uas-gitflgfp females. 22% of the pupal progeny carrying one copy of each of the dx and git transgenes and one copy of mesodermal driver hatched to adults. 77
% Genotypes Viability / uas-dx-gst/ 24bgal4/ 0 % (n=585) / uas-dx-gst/ 24bgal4-uas-gitfl-gfp/ (rescue) 21.3% (n=544) / uas-dx-gst/ 24bgal4-uas-gitshdc-gfp/ 0.6% (n=313) / uas-dx-gst/ 24bgal4-uas-gitshd -gfp / 3.9% (n=284) / uas-dx-gst/ 24bgal4-uas-gitc-gfp / 3.6% (n=212) n = total number of pupae counted Table 4.2A.1. GIT full length but not GIT domains significantly suppressed the pupal lethality of Dx-overexpression. None of the pupae hatched to adults when Dx was overexpressed under 24bgal4 mesodermal driver (/ uas-dx-gst / 24Bgal4 /) 21.3% of pupae hatched to adults when GIT was overexpressed together with Dx (/ uas-dx-gst / 24Bgal4-uas-gitflgfp /) no significant rescue of pupal lethality was obtained when individual GIT domains were used in the overexpression assay (<4% viability). Percentages were calculated as follows: number of hatched flies / total number of pupae x 100. 78
A B C D Figure 4.2A.2. Suppression of pupal lethality of Dx overexpression by GIT overexpression. dx-overexpression lethal phenotype under the influence of 24bgal4 mesodermal driver is simultaneously rescued by git-overexpression. Pupal lethality is partially rescued in progenies of the genotype / uas-dx-gst / 24bgal4-uasgitflgfp/ Panels A (normally hatched control wt fly), B (normally hatched rescued female fly of the genotype / uas-dx-gst / 24bgal4-uas-gitflgfp/) and C (a partially rescued / uas-dx-gst / 24bgal4-uas-gitflgfp/ female fly that partially hatched but remains stuck to its pupal case) are shown panel D is the same female fly shown in C after being forcibly removed from its pupal case. 79
I have also made use of available transgenic fly lines (generated by Bahri et al., unpublished) which carry different GIT domains (N-terminal, middle and C-terminal domains) on UAS-constructs. To express each of the different GIT domains in combination with the transgene uas-dx-gst, genetic crosses between homozygous males of the genotype /y uas-dx-gst / uas-dx-gst / and homozygous females of the genotype / / 24bgal4-uas-git-domain-gfp / 24bgal4-uas-git-domain-gfp were performed and their progeny carrying one copy of each transgene and one copy of driver were scored for pupal lethality (Table 4.2A.1). No significant rescue of pupal lethality (only 0.6-3.9%) was observed in this experiment, suggesting that all the domains of GIT have to be present in order to significantly rescue the toxic effects of Dx overexpression. The low percentage (0.6-3.9%) of adults hatching in the presence of constructs expressing different domains of GIT may be due to the possibility that some of the GAL4 in these animals would be used to drive uas-git-domain and consequently less of it is available to drive uas-dx in this scenario, a lower expression level of uas-dx would be expected. 4.2B. Embryonic muscle defects associated with Dx overexpression under mesodermal driver are suppressed by GIT overexpression The data presented in chapter 3 showed that mesodermal overexpression of Dx leads to muscle defects in the embryo. Muscles in embryos carrying one copy of uas-dx 80
and one copy of the uas-git transgene driven by the mesodermal driver were inspected and their phenotypes were scored and compared to embryos overexpressing the uas-dx transgene alone (Fig. 4.2B1 Table 4.2B.1). Expression of GIT together with Dx caused an evident reduction (from 8.7% to 1.6%) in the percentage of missing lateral transverse (LT) muscle phenotype that was associated with Dx overexpression alone. In addition, the muscle guidance phenotype of ventral oblique muscles associated with Dx overexpression (Fig. 4.2B.1B) was also considerably rescued by simultaneous expression of GIT (from 15% to 3.3% Table 4.2B.1 Fig. 4.2B.1C). The data suggests that overexpression of GIT may counteract the toxic effect of Dx overexpression in the embryonic mesoderm. 81
A A B C VM VM VO5 VM Figure 4.2B.1. Suppression of muscle phenotype associated with Dx overexpression by GIT overexpression. Ventral views of MHC-stained late stage embryos from wt control (A), Dx overexpressing (B, / uas-dx-gst / 24Bgal4 /) and rescued (C, / uas-dx-gst / 24Bgal4-uas-gitflgfp /) are shown mesodermal overexpression of GIT together with Dx (C) partially rescues the muscle guidance phenotype of ventral oblique muscles seen in overexpression of Dx alone (B, arrow). VM: ventral midline (arrowhead). All embryos are oriented anterior is to the left. VOs LTs VOs Genotypes Defective Guidance shape Missing defects / uas-dx-gst/ 24bgal4/ 4.3% (n=161) 8.7% (n=151) 15% (n=161) / uas-dx-gst / 24bgal4-uas-gitflgfp/ (rescued) 1.1% (n=183) 1.6% (n=183) 3.3% (n=183) n = number of hemisegments Table 4.2B.1. Suppression of muscle phenotypes associated with Dx overexpression by GIT overexpression. Percentages of embryonic muscles defects associated with mesodermal overexpression of Dx (/ uas-dx-gst / 24Bgal4 /) are decreased by simultaneous overexpression of GIT together with Dx (/ uas-dx-gst / 24Bgal4-uasgitflgfp /). Percentage of muscle defects were calculated as follows: number of hemisegments with muscle defects/total number of hemisegments x 100. 82
4.2C. Pupal lethality and wing notching phenotypes associated with Dxoverexression under epithelial driver are suppressed by overexpression of GIT In order to further characterize the interaction between dx and git in overexpression studies, a 32bgal4 epithelial driver was used this epithelial driver is known to expresses GAL4 in imaginal discs of larvae which give rise to adult structures such as wings later in development. To examine the phenotypes of Dx overexpression under the control of the epithelial driver, flies carrying two copies of uas-dx-gst transgene were crossed to flies carrying two copies of 32bgal4 driver and their progenies carrying one copy of transgene and one copy of driver were collected and scored for pupal lethality and adult phenotypes (Fig. 4.2C.1 Table 4.2C.1 Table 4.2C.2). In this overexpression experiment, 94.4% pupal lethality (pupal lethality here refers to unhatched and half-way hatched pupae) was observed in the progeny carrying one copy of transgene and one copy of driver (Table 4.2C.1). All flies that completely hatched from this experiment were females and no progeny males were survived to adulthood. In addition, the hatched females exhibited loss of tissues from their posterior-wing margin (Fig. 4.2C.1A). Simultaneous overexpression of GIT and Dx under the 32bgal4 epithelial driver partially rescued the pupal lethal phenotype observed with Dx overexpression alone (Table 4.2C.1). Furthermore, the wings of the completely hatched flies were also normal and did not show the loss of tissue phenotype that was observed in Dx overxpression 83
A B Figure 4.2C.1. Suppression of notched wing phenotype associated with epithelial overexpression of Dx by GIT overexpression. Panel A is a / uas-dx-gst /32bgal4 / female showing a notched wing phenotype on the posterior-wing margin (arrows) and panel B is a / uas-dx-gst / 32bgal4-uas-gitflgfp/ rescued female fly with a normal posterior wing margin (arrows). Genotypes / uas-dx-gst/ 32bgal4/ / uas-dx-gst / 32bgal4-uas-gitflgfp/ n = total number of pupae % Viability 5 % (n=402) 52.5% (n=463) Table 4.2C.1. Suppression of pupal lethality phenotype associated with Dx epithelial overexpression by GIT overexpression. The percentage of viable flies associated with epithelial overexpression of Dx (/ uas-dx-gst / 32bgal4 /) is significantly increased (from 5.6% to 52.5%) by simultaneous overexpression of GIT together with Dx (/ uas-dx-gst / 32bgal4-uas-gitflgfp /). Percentages were calculated as follows: number of hatched flies / total number of pupae x 100. 84
% Genotypes posterior wing margin defects / uas-dx-gst/ 32bgal4/ 93.3% (n=30) / uas-dx-gst/ 32bgal4-uas-gitflgfp/ (rescued) 0% (n=100) n = number of wings Table 4.2C.2. Suppression of notched wing phenotype associated with epithelial overexpression of Dx by GIT overexpression. Overexpression of Dx alone under 32bgal4 epithelial driver (/ uas-dx-gst /32bgal4 /) resulted in 93% of female wings showing notched (loss of tissue) phenotype at their posterior margin overexpression of GIT together with Dx (/ uas-dx-gst / 32bgal4-uas-gitflgfp/) completely rescued this phenotype. Percentages were calculated as follows: number of notched female wings / total number of female wings x 100. 85
alone (Fig. 4.2C.1 Table 4.2C.2). The data suggests that overexpression of GIT under the epithelial driver may antagonize the toxic effect of Dx overexpression, again similar to the suppression effect observed when overexpression of GIT and Dx was done using the 24bgal4 mesodermal driver (see section2 4.2A and 4.2B). 4.2D. Pupal lethality associated with dominant-negative Dx-ΔPRM overexpression under mesodermal driver is not suppressed by GIT overexpression The data described in sections 4.2A-4.2C suggested that Dx and GIT may have antagonistic effects on each others function in the overexpression assay. Based on this result, we deduced that if dominant-negative Dx-ΔPRM was used (instead of Dx) in the overexpression assay together with GIT, GIT would be expected to enhance the lethal phenotype of Dx-ΔPRM overexpression. To test this prediction, flies homozygous for uas-dx ΔPRM and flies homozygous for 24bgal,uas-gitflgfp were mated and the extent of pupal lethality in their progeny carrying one copy of driver and one copy of each of the transgenes was scored (Fig. 4.2D.1 Table 4.2D.1). The results in Table 4.2D.1 show that the overexpression of GIT together with Dx-ΔPRM did not significantly modify the pupal lethal phenotype associated with Dx-ΔPRM overexpression alone (85.3% as compared to 90.4% lethality with Dx alone). The data suggests that GIT does not interact with Dx- ΔPRM in the overexpression assay. A plausible deduction from this experiment is that the proline-rich motif (deleted in Dx-ΔPRM) may be required to mediate the antagonistic 86
effect of GIT on Dx. Alternatively, Dx-ΔPRM may be acting on new mesodermal targets (other than just endogenous Dx) which are not sensitive to the action of GIT. Overexpression of dominant negative dx ΔPRM in combination with git using 24bgal4 driver: 24bgal4-uas-gitfl-gfp 24bgal4-uas-gitfl-gfp X y uas- dx ΔPRM -gst uas- dx ΔPRM -gst & uas- dx ΔPRM -gst 24bgal4-uas-gitfl-gfp uas- dx ΔPRM -gst 24bgal4-uas-gitfl-gfp y 80% pupal lethal Figure 4.2D.1. Genetic interaction cross between homozygous uas-dx deltaprm -gst males and homozygous 24bgal4-uas-gitflgfp females. % Genotypes Lethality / uas-dx ΔPRM -gst / 24bgal4 / 91.4% (n=292) / uas-dx ΔPRM -gst / 24bgal4-uas-gitflgfp / 85.3% (n=476) n = total number of pupae counted Table 4.2D.1. GIT does not significantly suppress Dx-ΔPRM lethal phenotype in the mesodermal overexpression assay. Overexpression of Dx-ΔPRM alone (/ uas-dxgst / 24Bgal4 /) was 90.4% pupal lethal and overexpression of GIT together with Dx- ΔPRM (/ uas-dx-gst / 24Bgal4gitflgfp /) did not significantly modify this 87
phenotype (~ 5%). Percentages were calculated as follows: number of hatched flies / total number of pupae x 100. 4.3. Subcelular localization of Dx and GIT proteins in embryonic muscles are not interdependent The results described in chapter 3 showed that endogenous Dx is mainly localized to the cytoplasm in embryonic muscles with very weak localization at muscle attachment sites upon overexpression. Interestingly, one of the alleles of dx, dx 1, was also found to code for a protein that localizes primarily to muscle attachment sites (see chapter 3), suggesting that Dx may contain protein sequences that permit such localization. GIT has also been found to localize to muscle attachment sites (Bahri et al., unpublished). In order to determine whether Dx and GIT subcellular localizations in embryonic muscles are dependent on each other, homozygous mutant embryos from the dx 152 null allele and homozygous mutant embryos from the git 21C allele were double stained with anti-dx / anti-mhc and anti-git / anti-mhc respectively (Fig. 4.3.1). The staining results showed that GIT localization to muscle attachments was not affected in dx mutants (Fig. 4.3.1B) and that Dx cytoplasmic localization in muscles was also not affected in git mutants (Fig. 4.3.1E). The data indicates that GIT and Dx subcellular localizations in embryonic muscles are not dependent on each other. 88
A Anti-MHC B Anti-GIT C Merged * VLs dx 152 Anti-MHC Anti-Dx D E F Merged * VLs dgit 21c Figure 4.3.1. GIT and Dx localizations in embryonic muscles are not interdependent. Panels A-C show the ventro-lateral view of a late stage dx 152 mutant embryo double stained with anti-mhc (A, red) and anti-git (B, green) C is a merged image of A and B it shows that GIT at the muscle attachments (B, arrows) is not affected by the dx mutation. Panels D-F show the ventro-lateral view of a late stage git 21C mutant embryo double-stained with anti-mhc (D, red) and anti-dx (E, green) F is a merged image of D and E it shows that Dx cytoplasmic localization in the muscle (E, arrow) is not affected by the git mutation. All embryos are oriented anterior is to the left and dorsal is up. 89