Second-Site Noncomplementation Identifies Genomic Regions Required for Drosophila Nonmuscle Myosin Function During Morphogenesis

Corresponding author: Daniel P. Kiehart, Duke University Medical Center, Department of Cell Biology, Research Drive, 307 Nanaline Duke Bldg., Durham, NC 27710, d.kiehart{at}cellbio.duke.edu (E-mail).

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila is an ideal metazoan model system for analyzing the role of nonmuscle myosin-II (henceforth, myosin) during development. In Drosophila, myosin function is required for cytokinesis and morphogenesis driven by cell migration and/or cell shape changes during oogenesis, embryogenesis, larval development and pupal metamorphosis. The mechanisms that regulate myosin function and the supramolecular structures into which myosin incorporates have not been systematically characterized. The genetic screens described here identify genomic regions that uncover loci that facilitate myosin function. The nonmuscle myosin heavy chain is encoded by a single locus, zipper. Contiguous chromosomal deficiencies that represent approximately 70% of the euchromatic genome were screened for genetic interactions with two recessive lethal alleles of zipper in a second-site noncomplementation assay for the malformed phenotype. Malformation in the adult leg reflects aberrations in cell shape changes driven by myosin-based contraction during leg morphogenesis. Of the 158 deficiencies tested, 47 behaved as second-site noncomplementors of zipper. Two of the deficiencies are strong interactors, 17 are intermediate and 28 are weak. Finer genetic mapping reveals that mutations in cytoplasmic tropomyosin and viking (collagen IV) behave as second-site noncomplementors of zipper during leg morphogenesis and that zipper function requires a previously uncharacterized locus, E3.10/J3.8, for leg morphogenesis and viability.

THE elaboration of complex structures is a hallmark of metazoan development. This elaboration occurs incrementally, initiating with pattern formation and cell fate determination and proceeding through succeeding waves of cellular communications and inductions, cellular differentiation, and ultimately, the generation of mature forms via morphogenetic movements. Morphogenesis requires a link between signaling molecules and the structural molecules that carry out the actual work. Genetic, molecular, biochemical and experimental embryological analyses in many animal systems have expanded our understanding of the signaling events, but we are only beginning to elucidate the connection between signaling and downstream cell biological effects. Further, elucidation of the supramolecular structure of the cytoskeleton required for movement during morphogenesis is in its early stages.

Nonmuscle myosin-II (henceforth referred to as myosin) plays a key role in morphogenesis throughout phylogeny. It drives cell shape changes required for cytokinesis, tissue remodeling and cellular locomotion. Myosin heavy chain has been identified in Saccharomyces cerevisiae, and mutations in the MYO1 gene result in aberrations in cytokinesis and cell growth (WATTS et al. 1987 ; SWEENEY et al. 1991 ). In Dictyostelium discoideum, formation of the fruiting body requires myosin function (DE LOZANNE and SPUDICH 1987 ; KNECHT and LOOMIS 1987 ; MANSTEIN et al. 1989 ). Disruption of the myosin gene does not block the early stages of this morphogenetic process, but formation of the stalk does not proceed past the mound stage. Recent microscopic analyses suggest that myosin-based contraction may be important in generating cell shape changes during Dictyostelium morphogenesis (KNECHT and SHELDEN 1995 ; SHELDEN and KNECHT 1996 ). Although a function has not yet been demonstrated, myosin has been identified within the cilium of rat photoreceptor cells, a structure that is the site of new disk membrane morphogenesis (WILLIAMS et al. 1993 ). Finally, a recent analysis of the morphogenetic process of ventral enclosure of the hypodermis in Caenorhabditis elegans is suggestive of a role for myosin-based contraction (WILLIAMS-MASSON et al. 1997 ). Laser ablation and cytochalasin treatments suggest that movement of the ventral epidermis occurs via a two step mechanism: first, a quartet of cells migrate ventrally and this is followed by an actin-based purse string contraction at the leading edge of the epidermis, which brings the remainder of the cell sheet to the ventral midline.

Using the available genetic and molecular tools, the role of myosin during Drosophila morphogenesis has been particularly well characterized. Myosin function has been identified at many stages during the life cycle, including oogenesis, embryogenesis and adult metamorphosis. Single-copy genes encoding the myosin heavy chain [zipper (zip)], regulatory light chain [spaghetti-squash (sqh)] and essential light chain have been cloned (KIEHART et al. 1989 ; KARESS et al. 1991 ; EDWARDS et al. 1995 ). In embryos, myosin is subcellularly localized in many tissues undergoing cell shape change, including the cleavage furrows of the cellularizing embryo, the apical end of cells that form the ventral furrow during gastrulation and the leading edge of the epidermis during dorsal closure (YOUNG et al. 1991 ; YOUNG et al. 1993 ). Strong alleles of zip are embryonic lethal, and mutant embryos have defects in head involution, axon guidance and defects in epithelial sheet movement during dorsal closure (COTE et al. 1987 ; ZHAO et al. 1988 ; YOUNG et al. 1993 ; J. PEDERSON, J. MAHAFFEY and D. KIEHART, unpublished data). Mutations in sqh result in defective cytokinesis and growth of third instar larvae (KARESS et al. 1991 ; EDWARDS and KIEHART 1996 ). Experimental manipulation of sqh expression has revealed additional requirements for myosin function. These experiments include the generation of germline clones in the ovary (WHEATLEY et al. 1995 ) and stage-dependent depletion of myosin by removing flies carrying a heat shock-sqh rescue construct from heat shock (EDWARDS and KIEHART 1996 ). These studies reveal that myosin function is required during oogenesis for appropriate somatic follicle cell migrations and ingressions and nurse cell cytoplasmic transport. It is also required for nuclear migrations in cleavage stage embryos (WHEATLEY et al. 1995 ). During later stages of the life cycle, myosin is required for proper morphogenesis of the adult eye and leg (VON KALM et al. 1995 ; EDWARDS and KIEHART 1996 ). Presumably, these examples represent only a subset of morphogenetic processes that rely on myosin-based contraction.

Analysis of the Broad-Complex (Br-C ) suggests that it regulates myosin function during imaginal disc morphogenesis. Molecular genetic analysis shows that zinc-finger transcription factors encoded by Br-C are critical for imaginal disc morphogenesis (KISS et al. 1988 ; DIBELLO et al. 1991 ). A screen for enhancers of a Br-C family member, broad¹, identified several loci that function during leg imaginal disc morphogenesis (GOTWALS and FRISTROM 1991 ). Ebr is an EMS-induced enhancer of broad¹ (GOTWALS and FRISTROM 1991 ). Subsequent analysis utilizing a cosmid bearing a zipper transgene (YOUNG et al. 1993 ) verified that Ebr is a mutation in the myosin heavy chain locus (zip^Ebr; GOTWALS 1992 ). Defects in leg morphogenesis give rise to the malformed phenotype (Figure 1; GOTWALS and FRISTROM 1991 ). The malformed phenotype reflects aberrations in cell shape changes during morphogenesis in pupal leg imaginal discs (CONDIC et al. 1990 ; reviewed in VON KALM et al. 1995 ). The malformation ranges in severity from a small deformation in the femur (Figure 1B) to extreme twisting and gnarling of the femur and tibia (Figure 1C). The genetic behavior of myosin, and the observation that myosin is subcellularly localized during leg elongation and during additional morphogenetic events (YOUNG et al. 1991 ; VON KALM et al. 1995 ), strongly support the hypothesis that myosin-based contraction drives these cell shape changes.

View larger version (136K): [in this window] [in a new window] [Download PPT slide]   Figure 1. —Malformed phenotype. Defects in cell shape changes driven by myosin-based contraction during pupal leg morphogenesis give rise to the malformed phenotype (mlf) (CONDIC et al. 1990 ; VON KALM et al. 1995 ). The greatest degree of cell shape change occurs within those cells that will compose the femur and tibia, and the third leg pair is most sensitive to perturbation. (A) In a wild-type adult, the femur and tibia are elongated and slender structures. (B) A weakly mlf leg is evident in a Df(2R)M41A +/+ zipEbr double heterozygous fly. The dent in the femur (open arrow) is characteristic of mlf, and there is a slight thickening of the tibia. (C) In severely mlf legs observed in a Df(2R)Jp8 +/+ zipEbr adults, the structure of the leg is grossly affected. The femur is gnarled and twisted upon itself (open arrow) and the tibia is also shortened and twisted. The presence of weakly mlf legs was considered diagnostic in these screens.

Genetic modifier screens have been used to identify a series of gene products required in a number of functional processes. For example, such screens have proven particularly powerful in identifying gene products that function during Drosophila eye development. An example of one such screen is exemplified by a study which utilized a temperature-sensitive mutation in sevenless, sev^B4 (SIMON et al. 1991 ). These flies are exquisitely sensitive to changes in function of other gene products required in the sevenless pathway. In the screen, EMS-induced mutations were tested for genetic interactions with sev^B4, just below its restrictive temperature. Seven E(sev) loci were identified in this screen and included mutations in Ras, Son of Sevenless, Hsp83 and cdc37 (SIMON et al. 1991 ; CUTFORTH and RUBIN 1994 ). In addition to conducting genetic modifier screens with EMS-induced mutations, a series of contiguous chromosomal deficiencies are available from the Bloomington Stock Center for systematic screening. Chromosomal deficiencies have been successfully utilized to identify genomic regions required for such divergent processes as DNA replication during embryogenesis (SMITH et al. 1993 ), embryonic midgut morphogenesis (BILDER and SCOTT 1995 ) and germline sex determination and differentiation (PAULI et al. 1995 ). While mapping the location of the Ebr locus, GOTWALS and FRISTROM 1991 identified two chromosomal deficiencies that behave as second-site noncomplementors of zip during imaginal disc leg elongation. Their observations indicated that a systematic screen to identify modifiers of zip function could be developed utilizing chromosomal deficiencies.

In this study, we have utilized second-site noncomplementation (SSNC) screens utilizing chromosomal deficiencies to identify genomic regions required for zip function. These screens assay for the malformed phenotype (mlf ). This approach was chosen because: (1) it is an F1 adult viable screen and mlf is readily scorable with a dissecting scope, (2) utilization of the chromosomal deficiencies allow screening of ~70% of the euchromatic genome in a relatively small number of crosses, and (3) redundant usage of gene products throughout development makes it likely that genes identified in this screen will function in additional morphogenetic processes. A total of 158 deficiencies on the X, second and third chromosomes have been tested. Forty-seven deficiencies interact genetically with zip, resulting in variable penetrance of mlf. Analysis of the overlap between these deficiencies and the behavior of the genetic interactions suggests there are at least 39 SSNC loci uncovered in these screens. Mutations in several genes uncovered by SSNC deficiencies, including cytoplasmic tropomyosin, viking [collagen 2(IV)] and a previously identified but uncharacterized locus, E3.10/J3.8, also behave as a SSNC of zip. This finer genetic mapping and the identification of cTm as a SSNC validates the efficacy of this screening approach. Since myosin function is so well conserved throughout phylogeny, it is likely that gene products identified in these screens will function similarly in other animals.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Stocks:
The Deficiency Kit stocks and l(2)01209 were obtained from the Bloomington Stock Center (Bloomington, IN). Cytology is based on LINDSLEY and ZIMM 1992 and FlyBase (http://cbbridges.harvard.edu.7081/). The zip^6.1 allele was from the KIEHART lab (YOUNG et al. 1993 ), and zip^Ebr (GOTWALS and FRISTROM 1991 ) was obtained from JIM FRISTROM (University of California, Berkeley). BILL SAXTON (Indiana University) provided khc ⁸, D2.33, D8.13, E3.10, E3.27 and E6.17 (SAXTON et al. 1991 ), and RUTH STEWARD (Rutgers University) provided J3.8. vkg ^BLK, vkg ^ICO and vkg ^SAK were obtained from DEBORAH KIMBRELL (University of Houston; RODRIGUEZ et al. 1996 ). E74^P[neo] and E74^DL-1 were obtained from CARL THUMMEL (University of Utah; FLETCHER et al. 1995 ). cTm^eg9 and cTm^er4 were obtained from ANNE EPHRUSSI (EMBL, Heidelberg; ERDELYI et al. 1995 ).

Second-site noncomplementation screen:
For the autosomal deficiencies, four to five zip^Ebr/SM5 or zip^6.1/CyO virgin females were mated to three to four Deficiency/Balancer males in 4–5 day broods at either 18° or 25°. At least 20 adult flies of the double heterozygous zip; Deficiency genotype (e.g., zip/+; Df/+) were scored for the malformed phenotype (mlf ) (GOTWALS and FRISTROM 1991 ). A similar mating scheme was used for the X chromosome deficiencies, except virgin females bearing the X chromosome deficiencies were mated with zip males and only adult female progeny were scored. The penetrance of mlf in the double heterozygotes was compared directly to the penetrance in their siblings heterozygous for the deficiency (e.g., Balancer/+; Df/+). A penetrance of mlf greater than 10% but less than 25% in the double heterozygotes and not attributable to the deficiency alone was classified as weakly interacting. Intermediate second-site noncomplementing deficiencies show a penetrance of mlf between 25–75% and strong interactors have a penetrance in excess of 75%.

Leg mounting:
The malformed phenotype was scored under the dissecting scope. For photography, malformed adults were preserved in 70% ethanol. Individual legs were dissected and mounted in CMPC10 (PolySciences Inc., Warrington, PA). The mounted legs were observed by brightfield microscopy using a 5x Neofluar objective (0.15 NA) on a Zeiss Axioplan microscope. Images were captured with a Hamamatsu CCD camera and Metamorph software. Contrast was adjusted using Adobe Photoshop software and the images were then exported to a Canvas file for montaging and labeling (KIEHART et al. 1994 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Pilot screens:
Pilot screens with previously identified, EMS-induced second-site noncomplementors (SSNCs) of zip^Ebr showed that strong, embryonic lethal zip alleles, such as zip², did not give rise to the malformed phenotype in double heterozygous mutant adults (J. FRISTROM, S. HALSELL and D. KIEHART, unpublished results). Instead, the post-embryonic lethal alleles, zip^2.1 and zip^6.1 (YOUNG et al. 1993 ), are sensitive to such genetic interactions (data not shown). In the assay for the malformed phenotype, the zip^2.1 and zip^6.1 alleles behave with similar characteristics. In the pilot screens, zip^Ebr is the most sensitive allele available for testing for second-site noncomplementation (data not shown), and this is consistent with the observations found in the deficiency screens described here (see below). Based on these observations, zip^Ebr and zip^6.1 were used in the second-site noncomplementation screens described here. The choice of zip^6.1 instead of zip^2.1 was arbitrary.

Malformed flies are observed at low levels in the zip^Ebr stock. This is not the case with the zip^6.1 stock. In order to determine the background level of mlf in flies heterozygous for zip^Ebr, zip^Ebr/SM5 virgins were crossed to either Oregon-R wild-type males or to males carrying a CyO balancer chromosome. In flies outcrossed to Oregon-R wild type (zip^Ebr/+), malformed legs are seen in 53/848 adult progeny (6% mlf ). In trans to a CyO balancer chromosome (zip^Ebr/CyO), 3% of the adults are malformed (2626 flies). Based on these observations, the genetic interactions observed between zip and the chromosomal deficiencies were arbitrarily classified as weak (mlf penetrance 10–24%), intermediate (mlf penetrance 25–75%) or strong (mlf penetrance >75%).

Deficiency screens:
The Bloomington Stock Center Deficiency Kits were used for F1 adult viable, second-site noncomplementation screens in order to identify genomic regions required for zipper (zip) function. zipper encodes the Drosophila nonmuscle myosin-II heavy chain gene (YOUNG et al. 1993 ). A total of 163 contiguous deficiency stocks are available for the X, second and third chromosomes, and they uncover approximately 70% of the euchromatic genome (FlyBase). Excluding exceptionally weak stocks and those that uncover zipper, 158 of the deficiency stocks were tested for genetic interactions with zip^Ebr and zip^6.1. Second-site noncomplementation of zip is indicated by the incidence of the malformed phenotype (mlf ) in the adult leg (Figure 1; GOTWALS and FRISTROM 1991 ). Initially, at least 20 adult progeny double heterozygous for zip and the tested deficiency were scored. In those cases where at least 10% of the flies were mlf, the cross was repeated and at least 50 additional flies were scored. Alternatively, those deficiencies that did not show an interaction with zip but overlap deficiencies that did show an interaction, were rescreened to confirm this observation.

Of the 158 tested stocks, 47 deficiencies, when double heterozygous with at least one of the tested zip alleles, give rise to mlf adults with a penetrance of at least 10% (Table 1 Table 2 Table 3). Expressivity of mlf may be weak or strong (cf. Figure 1B TO C), but it is not necessarily correlated with the degree of penetrance. A weak mlf phenotype is considered diagnostic. Generally, SSNC is observed with greater penetrance with zip^Ebr than with zip^6.1. Of those deficiencies tested at 25° and 18°, 22 show a higher penetrance of mlf at 25°, 8 are more severe at 18° and 13 show no significant temperature sensitivity. Analysis of the relationship between these deficiencies and their genetic behavior suggests that a minimum of 39 loci behave as second-site noncomplementors (SSNC) of zipper. The genetic interactions observed between these deficiencies and zip are detailed below.

X Chromosome
There are 42 deficiency stocks available in the Deficiency Kit for the X chromosome. The data for the SSNC deficiencies are shown in Table 1 and Figure 2, and those deficiencies which fail to interact in these screens are shown in Table 10. A total of 14 deficiencies on the X chromosome behave as SSNCs of zip, six with an intermediate penetrance (25–75% penetrance), and the remaining eight as weak interactors (penetrance of 10–24%). There are no strong interacting deficiencies on the X chromosome.

View larger version (12K): [in this window] [in a new window] [Download PPT slide]   Figure 2. —Second-site noncomplementing deficiencies on the X chromosome. Of the tested X chromosome linked deficiencies, 14 behave as SSNCs of zip. This represents a minimum of 11 loci. Six of the deficiencies act as intermediate SSNCs (green boxes), and eight are weak SSNCs (yellow boxes; Table 1). Tested, noninteracting deficiencies are indicated by the white boxes (Table 10). Of the available deficiencies, no strong SSNC deficiencies were identified on the X chromosome. The chromosome map is modified from LEFEVRE 1976 .

Intermediate, second-site noncomplementing X chromosome deficiencies: 4C15-16; 4F5: Overlapping deficiencies Df(1)A113 and Df(1)JC70 interact with zip^Ebr but not zip^6.1. At 18°, Df(1)JC70 ; zip double heterozygotes exhibit mlf with a penetrance of 43% and a similar penetrance of 37% at 25°. The penetrance of mlf with Df(1)A113 is lower (23% mlf), but this may be an underestimation because the stock carries a duplication for region 3C2; 5A1-2 (Dp(1;2)w[+]) that cannot be followed phenotypically; thus, the scored class includes flies that carry this duplication. However, this duplication cannot be responsible for the appearance of mlf because 50% of the zip heterozygotes (+/+; zip^Ebr/+) or deficiency heterozygotes (Df/+; +/+) would also be heterozygous for the duplication, but mlf is not seen in these flies. Neither deficiency interacts with zip^6.1 (2–3% mlf ). The smallest overlap between these regions is cytologically 4C15-16; 4F5, although the higher penetrance with Df(1)JC70 indicates that it might uncover a second locus. Alternatively, these deficiencies may uncover the same locus, but the duplication in the Df(1)A113 stock masks the full penetrance of mlf in these flies.

7D10; 8A4-5: Df(1)RA2 uncovers the cytogenetic interval 7D10; 8A4-5 and acts as a SSNC of intermediate penetrance with zip^Ebr at 25° (32% mlf ) and a weaker penetrance at 18° (15% mlf ). There is no observed interaction with zip^6.1 (6% mlf ). An overlapping deficiency, Df(1)KA14 does not interact genetically with zip (7% mlf ). Based on the cytology, the interacting locus is confined to 7D10; 7F1-2.

9B1-2; 11A1-2: Cytogenetic interval 9B1-2; 11A1-2 is uncovered by three overlapping deficiencies, Df(1)v-L15, Df(1)v-N48 and Df(1)HA85, all of which behave as intermediate SSNCs. Each deficiency displays a slightly different interaction with zip. Df(1)v-L15 shows a penetrance of mlf of approximately 40% with zip^Ebr at both temperatures, and a weak interaction with zip^6.1 at 25° (10% mlf ). Df(1)v-N48 overlaps Df(1)v-L15 between 9F; 10A1-2, but shows a slightly weaker penetrance with zip^Ebr (29% mlf at 25°) and no interaction with zip^6.1 (2% mlf ). This may reflect either genetic background differences in the stocks or it may indicate that Df(1)v-L15 uncovers an interacting locus not uncovered by Df(1)v-N48. Given that Df(1)C52 does not interact with zip (6% mlf ), this second possible SSNC locus is confined to 9C-D; 9F. More proximally, Df(1)HA85 also acts as an intermediate SSNC with zip^Ebr (38%) and a weak SSNC with zip^6.1 (20% mlf ). Again, this genetic behavior differs from Df(1)v-N48, and may reflect the presence of a SSNC locus in region 10C3-5; 11A1-2. Operationally, the overlap between these three deficiencies defines a minimum number of interacting loci as two, but the differences observed in their interactions with the zip alleles indicate that there may be three or more separate loci.

14C2-4; 15B2-C1: In cytogenetic interval 14C2-4; 15B2-C1, a single deficiency, Df(1)r-D1, interacts with zip^Ebr to give a mlf penetrance of 29% at 25°. This stock is heterozygous for Dp(1;4)r[+], an unmarked duplication covering interval 14A1-2; 16A7-B1, and this duplication may affect the occurrence of mlf flies. However, the presence of the duplication does not result in mlf in the zip heterozygous (+/+; zip^Ebr/+) or deficiency heterozygous (Df/+; +/+) siblings (see above, 4C15-16; 4F5). Based on these observations, it is likely that the penetrance observed in the Df(1)r-D1; zip^Ebr double heterozygotes is an underestimate.

Weak, second-site noncomplementing X chromosome deficiencies: 1B; 3C2: Cytological region 1B; 3C2 contains a weakly interacting SSNC. Two deficiencies within this interval, Df(1)sc-J4 and Df(1)64c18 interact with zip^Ebr with a weak penetrance of mlf (11% and 12% mlf, respectively). Deficiencies Df(1)BA1 and Df(1)JC19 do not interact genetically with either zip allele (3% mlf); therefore, the interacting locus may be confined to cytological region 2E1-2; 2F6. Alternatively, each deficiency may uncover separate loci, one confined to 2A; 2E1-2 and a second to 3A3; 3C2. The lack of interaction seen with Df(1)JC19 could result if a suppressor(s) of mlf lies on the chromosome outside the deficiency. It is more likely that two loci in the region act as SSNCs of zip because the Broad-Complex maps to 2B5, which is uncovered by Df(1)sc-J4 but not by Df(1)64c18 . A deficiency specific for the Broad-Complex, npr 1, is also a weak SSNC of zip^Ebr (GOTWALS and FRISTROM 1991 ). This genetic behavior is similar to that observed with Df(1)sc-J4.

5E3-8; 6B: A single deficiency, Df(1)G4e[L]h24i[R] uncovers region 5E3-8; 6B and interacts with zip^6.1 (19% mlf ) but not with zip^Ebr (5% mlf ). This is only one of two cases in which zip^6.1 exhibits a higher penetrance of mlf than zip^Ebr. The interaction with zip^6.1 is weak and occurs at 25°. Df(1)5D and Df(1)JF5 do not interact with either zip allele, therefore, the smallest interacting interval is 5E8; 6B.

7B2-4; 7C1: Weak SSNC is observed with overlapping deficiencies Df(1)ct-J4 and Df(1)ct4b1 and zip^Ebr (23% and 20% mlf, respectively). Temperature sensitivity is not observed in either case. The smallest interval uncovered by these deficiencies is 7B2-4; 7C1.

12A; 12E and 18E1-2; 20: Two additional genetic regions are defined by single deficiencies which act as weak SSNCs. Df(1)g uncovers 12A; 12E and at 25°, interacts with zip^Ebr with a penetrance of 24%. Overlapping deficiency Df(1)RK2 does not interact (3% mlf ), so the smallest defined interval is 12A; 12D2. At 25°, Df(1)HF396 acts as a weak SSNC with zip^Ebr (22% mlf ) and zip^6.1 (17% mlf ). Noninteracting deficiencies Df(1)DCB1-35b (1% mlf ) and Df(1)A209 (4% mlf ) further subdivide this interval to 18E1-2; 19F1-2.

Two genes, regulatory light chain (spaghetti-squash; KARESS et al. 1991 ) and essential light chain (EDWARDS et al. 1995 ), are critical for myosin function and map to the X chromosome. None of the deficiencies available in the deficiency kit uncover the essential light chain, but two deficiencies, Df(1)N73 and Df(1)5D, should uncover spaghetti-squash (sqh). Neither deficiency, however, interacts with zip (3% mlf). Therefore, a reduction in the gene dosage of sqh is not sufficient to elicit the malformed phenotype in combination with zip.

Second Chromosome
Fifty-six deficiency stocks on the second chromosome are available for analysis. Twenty-one deficiencies on the second chromosome behave as SSNCs of zip (Figure 3; Table 2; Table 11). This represents a greater proportion of SSNCs deficiencies on the second chromosome than those available on the X or third chromosomes. One of the deficiencies acts as a strong SSNC nine exhibit an intermediate penetrance of mlf, and the remaining 11 are weak SSNCs.

View larger version (24K): [in this window] [in a new window] [Download PPT slide]   Figure 3. —Second-site noncomplementing deficiencies on the second chromosome. Of the tested second chromosome-linked deficiencies, 21 behave as SSNCs of zip. This represents a minimum of 17 loci. One deficiency, Df(2R)Jp8, behaves as a strong SSNC deficiency (red box), nine of the deficiencies act as intermediate SSNCs (green boxes), and 11 are weak SSNCs (yellow boxes; Table 2). Tested, noninteracting deficiencies are indicated by the white boxes (Table 11). The chromosome map is modified from LEFEVRE 1976 .

Strong, second-site noncomplementing second chromosome deficiency: 52F5-9; 52F10-53A1: On the right arm of the second chromosome, within cytological interval 52F5-9; 52F10-53A1, lies a locus that is a strong SSNC of zip. At 25°, 98% of Df(2R)Jp8 +/+ zip^Ebr flies are malformed and the penetrance is 76% in trans to zip^6.1. In addition to displaying mlf with a high penetrance, Df(2R)Jp8 is semilethal in combination with zip^Ebr (Table 4). Df(2R)Jp8 +/+ zip^Ebr flies are recovered at 55% of the expected Mendelian frequency. Df(2R)Jp8 uncovers the kinesin heavy chain (khc) locus (SAXTON et al. 1991 ). We tested a strong kinesin heavy chain mutation, khc ⁸, and six additional EMS mutations uncovered by Df(2R)Jp8 for genetic interaction with zip^Ebr at 25° (Table 4; R. STEWARD, personal communication; B. SAXTON, personal communication; SAXTON et al. 1991 ). The khc⁸ allele does not interact with zip^Ebr (<1% mlf ) nor do four of the six other EMS mutations. Mutations E3.10 and J3.8, however, do interact with zip^Ebr with characteristics similar to those observed with Df(2R)Jp8. Flies double heterozygous for zip^Ebr and E3.10 are semi-viable (16% of expected Mendelian frequency) and 100% of the escaper adults exhibit mlf. One hundred percent of J3.8 +/+ zip ^Ebr double heterozygous flies are mlf, and are observed with only 53% of the expected Mendelian frequency. E3.10 and J3.8 are recessive lethals that fail to complement each other (data not shown; R. STEWARD, personal communication) Taken together, it is likely that the genetic interaction seen with Df(2R)Jp8 is specific to the interval uncovered by the deficiency and that E3.l0 and J3.8 are mutations in the locus responsible for this interaction. Further characterization of E3.10/J3.8 is underway.

51C3; 52F5-9: Abutting the interval uncovered by Df(2R)Jp8 is a second deficiency, Df(2R)Jp1, that behaves as an intermediate SSNC of zip^Ebr. This deficiency uncovers a locus other than E3.10/J3.8 based on the following: First, the interaction seen with Df(2R)Jp1 is specific to zip^Ebr and penetrance of mlf is lower (40%) and is seen only at 25°. Also, no reduction in viability is seen in these flies (data not shown). Finally, the mutagenesis that generated E3.10 was designed such that the mutation is uncovered by Df(2R)Jp8 but not uncovered by Df(2R)Jp1 (SAXTON et al. 1991 ). Therefore, an additional SSNC locus of zip falls in interval 51C3; 52F5-9. The myosin light chain kinase maps to this interval (KOJIMA et al. 1996 ; TOHTONG et al. 1997 ; M. CHAMPAGNE, K. EDWARDS and D. KIEHART, unpublished results), but mutations in the gene are not currently available for analysis.

Intermediate, second-site noncomplementing second chromosome deficiencies: 25A4-5; 26B2-5: An intermediate penetrance of 30% of mlf is seen between Df(2L)c1-h3 and zip^Ebr. This deficiency uncovers interval 25D2-4; 26B2-5. Df(2L)sc19-5 overlaps this interval between 25D2-4; 25D5-7, and it behaves as a weak SSNC. Based on its cytological location, viking [vkg, encoding collagen 2(IV); YASOTHORNSRIKUL et al. 1997 ] should be uncovered by Df(2L)sc19-5 but not by Df(2L)c1-h3. Mutations in the vkg locus have been tested for second-site noncomplementation with zip ^Ebr (Table 5). l(2)01209 is a P-element insertion that fails to complement vkg (data not shown). In trans to zip ^Ebr, 15% of the l(2)01209 heterozygous flies are mlf. Additionally, vkg ^ICO acts as a weak SSNC of zip (10% mlf ), although two additional alleles, vkg ^BLK and vkg ^SAK (6% and 5% mlf, respectively) do not interact with zip^Ebr by our criteria. The differences in SSNC behavior of these vkg alleles with zip may reflect differences in their strength, although previous complementation analysis does not address this issue directly (RODRIGUEZ et al. 1996 ).

29F7-30A1; 30C2-5: Cytogenetic interval 29F7-30A1; 30C2-5 is uncovered by Df(2L)30A-C. This deficiency is a SSNC of zip^Ebr, with a phenotypic penetrance of 26%, but does not interact with zip^6.1 (2% mlf ). The genetic interaction is only seen at 25°. No additional deficiencies within the Deficiency Kits overlap this interval, so the focus of the interaction has not been mapped any more finely.

36A8-9; 36E1-2 and 36E4-F1; 38A6-7: Two adjacent deficiencies, Df(2L)H2O and Df(2L)TW50, span 36A8-9; 36E1-2 and 36E4-F1; 38A6-7, respectively. Each shows an intermediate penetrance of mlf of 28% with zip^Ebr at 25°. This reveals at least two SSNCs loci. Df(2L)TW137 (6% mlf ) is noninteracting and overlaps each of these deficiencies, and noninteracting Df(2L)TW84 (0%) overlaps the proximal end of Df(2L)TW50. Thus, at least one interacting locus is confined to 36A8-9; 36C2-4 and the other is restricted to 37B9-C1; 37F5-38A1.

41A; 42B1-3: Df(2R)nap1 maps cytogenetically to 41D2-E1; 42B1-3. It is an apparent strong SSNC of both zip^Ebr and zip^6.1. At 25°, mlf is seen in 82% of flies double heterozygous for the deficiency and zip^Ebr and in 75% of the flies carrying zip^6.1. A caveat to the interpretation of this data is that the deficiency chromosome shows dominance for mlf since flies that are singly heterozygous for the Df(2R)nap1 also display mlf (Table 6). The appearance of mlf in the Df(2R)nap1 heterozygotes is dependent on genetic background. In crosses to zip^Ebr/SM5, mlf is seen in approximately 45% of the Df(2R)nap1/SM5 flies (Table 6), while the penetrance is 65–82% in Df(2R)nap1/CyO flies (data not shown). If the penetrance of mlf in the case of the zip^Ebr is strictly additive, then the penetrance of mlf in the Df(2R)nap1 +/+ zip^Ebr double heterozygotes would be approximately 40% at 25°. It is unlikely that the genetic interaction seen with zip^Ebr and Df(2R)nap1 is a manifestation of only the genetic background of the deficiency stock, because an overlapping deficiency present in chromosomal rearrangement In(2R)bw[VDe2L]Cy[R] (41A-B; 42A2-3) also behaves as an intermediate strength SSNC of zip^Ebr, exhibiting mlf in 43% of the flies at 25° (Table 2 and Table 6). Flies heterozygous for In(2R)bw[VDe2L]Cy[R] also exhibit mlf in the absence of zip (10–12%; Table 6) but with considerably lower penetrance than Df(2R)nap1. Based on cytology, one of the loci encoding cytoplasmic actin (Act42A) maps to 42A (FYRBERG et al. 1981 ), but it is unlikely to be the SSNC because Df(1)nap9 (42A1-2; 42E6-F1) should also uncover Act42A but is not a SSNC (Table 11). Assessing the data observed with these three stocks shows at least one interacting locus falls within 41D2-E1; 42A1-2. Additionally, Df(2R)M41A4 uncovers 41A and acts as a weak SSNC. Thus, cytogenetic interval 41A-F contains at least two loci that interact with zip.

59D5-9; 60D: The tip of the right arm of the second chromosome is represented by several SSNC deficiencies that do not uncover zip, verified by their ability to complement the two zip alleles, but do uncover a number of interacting loci. Flies heterozygous for Df(2R)bw[VDe2L]Px[Kr] have blistered wings. In addition to the blistered wings, flies double heterozygous for the deficiency and either zip^Ebr or zip^6.1 display mlf with a penetrance ranging from 35–41% and no temperature sensitivity. Df(2R)or-BR6 overlaps the proximal region of Df(2R)bw[VDe2L]Px[Kr] between 59D5-9; 60B3-8. Unlike Df(2R)bw[VDe2L]Px[Kr], Df(2R)or-BR6 interacts with zip^Ebr and zip^6.1 only at 18° and results in a weak penetrance of mlf (12–15%). Distally, the deficiency uncovered by In(2LR)Px[4] overlaps Df(2R)bw[VDe2L]Px[Kr] between 60B; 60D1-2. In(2LR)Px[4] is an intermediate SSNC of zip^Ebr and zip^6.1 at 18° (56% and 24% mlf, respectively), consistent with the results seen previously by GOTWALS and FRISTROM 1991 with zip^Ebr. Taken together, these results suggest there are at least two interacting loci uncovered by these deficiencies. One locus interacts specifically at 18° and would fall into 60B1-8, the region overlapped by Df(2R)or-BR6 and In(2LR)Px[4]. At least one additional locus would fall in interval 60B3-8; 60D. GOTWALS and FRISTROM 1991 identified two loci that interact with zip^Ebr in this interval: bs and l(2)B485. Contrary to their observations, an interaction was not observed between Df(2R)PX2 and zip.

Weak, second-site noncomplementing second chromosome deficiencies: In addition to the three weakly interacting deficiencies on the second chromosome described above, there are eight additional deficiencies, each defining a separate genomic region, that are also weakly interacting. On the left arm of the second chromosome, there are four additional weak SSNC deficiencies in addition to Df(2L)sc19-5.

23C3-5; 23D1: Df(2L)JS32 uncovers interval 23C3-5; 23D1 behaves as a weak SSNC (11% mlf ).

28E4-7; 29B2-C1: Genomic region 28E4-7; 29B2-C1 is uncovered by Df(2L)TE29Aa-11 which interacts weakly with zip^Ebr at 18° and 25° (15–16% mlf) and also with zip^6.1 at 18° (11% mlf ).

31A; 32C-E: Df(2L)J39 uncovers 31A; 32C-E and also interacts with zip^Ebr and zip^6.1 (13–14% mlf ).

35B1-3; 35E6: In cytological location 35, Df(2L)osp29 uncovers a locus that behaves as a weak SSNC of zip^Ebr at 18° (11% mlf ). Df(2L)b87e25 and Df(2L)r10 do not interact with zip (1% and 6% mlf, respectively); therefore this locus is restricted to 35B10-C1; 35E1.

On the right arm of the second chromosome, there are four weak SSNC deficiencies in addition to Df(2R)or-BR6.

46A; 46C: Df(2R)B5 uncovers interval 46A; 46C. At 25°, Df(2R)B5; zip^Ebr double heterozygous flies exhibit mlf at 11% penetrance.

49A4-13; 50C23-D2: In the genomic region 49A4-13; 50C23-D2, two overlapping deficiencies, Df(2R)vg-C and Df(2R)CX1, act as weak SSNCs of zip^Ebr, although they behave with slightly different characteristics. Df(2R)vg-C shows the interaction only at 18° (12% mlf ) while the interaction shows both a higher penetrance and no temperature sensitivity in the case of Df(2R)CX1 (19–23% mlf ). Either each deficiency uncovers interacting loci within the nonoverlapping regions, or differences in the genetic background of these stocks affect the specificity of the interactions.

55A; 55F: Finally, Df(2R)PC4 uncovers interval 55A; 55F, and it interacts specifically with zip^Ebr but not zip^6.1 with a penetrance of 20–24%. The more proximal deficiencies Df(2R)Pc17B and Df(2R)Pc111B (4% and 1% mlf ) do not interact, therefore the interacting locus is mapped to 55C1-3; 55F.

Third Chromosome
Of the 65 deficiencies available on the third chromosome, 12 behave as SSNCs of zip (Figure 4; Table 3; Table 12). One deficiency behaves as a strong SSNC and in combination with zip^Ebr shows reduced viability. Two deficiencies behave as intermediate SSNCs and the remainder interact weakly. A minimum of 11 interacting loci can be inferred.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 4. —Second-site noncomplementing deficiencies on the third chromosome. Of the tested third chromosome-linked deficiencies, 12 behave as SSNCs of zip. This represents a minimum of 11 loci. One deficiency, Df(3L)81k19, behaves as a strong SSNC deficiency (red box), two of the deficiencies act as intermediate SSNCs (green boxes), and nine are weak SSNCs (yellow boxes; Table 3). Tested, noninteracting deficiencies are indicated by the white boxes (Table 12). The chromosome map is modified from LEFEVRE 1976 .

Strong, second-site noncomplementing third chromosome deficiency: 73A3; 74F: In the presence of zip^Ebr, Df(3L)81k19 (cytogenetic interval 73A3; 74F) behaves as a strong SSNC at both 18° and 25°, exhibiting a penetrance of mlf in 77–84% of the flies. There is a decrease in viability of double heterozygous mutant flies, as compared to their SM5/+; Df(3L)81k19/+ siblings (53% of expected Mendelian frequency; Table 7). At 25°, flies with the deficiency in trans to zip^6.1 show a 17% penetrance of mlf. The locus encoding an ecdysone-responsive ets domain transcription factor, E74, lies within this interval (BURTIS et al. 1990 ). Circumstantial evidence that favors E74 as a SSNC is as follows. First, E74 is uncovered by Df(3L)81k19 (data not shown; FLETCHER et al. 1995 ). Second, it is expressed at late third larval instar (THUMMEL et al. 1990 ), coincident with the initiation of the major cell shape changes that occur during the morphogenesis of the leg (CONDIC et al. 1990 ; FRISTROM and FRISTROM 1993 ). Finally, it has been shown to interact genetically with the Broad-Complex (FLETCHER and THUMMEL 1995 ), members of which are also known to interact genetically with zip (GOTWALS and FRISTROM 1991 ). E74 is a complex locus, encoding two overlapping transcripts with separate promoters, and two mutations, E74^P[neo] and E74^DL-1, affect each transcript independently and complement each other genetically (BURTIS et al. 1990 ; FLETCHER et al. 1995 ). Because E74 appeared to be a good candidate for acting as the SSNC, we tested it, but found it did not interact genetically with zip. Each E74 mutation was tested with zip for second-site noncomplementation (Table 7). For each allele independently, no interaction was seen with zip^Ebr (0% mlf). Df(3L)81k19, however, should remove both E74 transcripts. In order to eliminate both transcripts from the flies, a double heterozygous stock was constructed, zip^Ebr/SM5; E74P ^[neo]/TM6C, and males from this stock were crossed to E74^DL-1/TM6C virgin females. The malformed phenotype was then assessed in flies heterozygous for zip^Ebr and transheterozygous for both E74 alleles. Again, no malformed flies were observed (0% mlf). Therefore, it is clear that E74 does not interact genetically with zip.

Since Df(3L)81k19 uncovers a strong SSNC of zip, finer genetic mapping was pursued (Figure 5; Table 7). Reported deficiencies that overlap Df(3L)81k19 are confined to the proximal side of the deficiency. A weakly interacting genetic region has been identified between 73A3-4; 73C1. Df(3L)st-bll, Df(3L)st7 and Df(3L)st4 interact with zip^Ebr, giving rise to mlf flies in 13–16% of the cases. Df(3L)st-f13 defines the proximal limit of this region because it does not interact (3% mlf). The low penetrance of mlf observed with these deficiencies suggests that at least one other SSNC locus of zip, falls within the interval 73C1; 74F, and perhaps, based on the distal breakpoint in Df(3L)st7, the locus is actually confined to 74A3-74F. Unfortunately, no published deficiencies subdivide this region. Eighteen lethal P-element insertions collected by the Berkeley Drosophila Genome Project (University of California, Berkeley, CA) fall within 73A-74F have been tested for SSNC of zip, but none of them interact (data not shown). In sum, we have yet to identify the locus or loci responsible for the strong SSNC observed when the Df(3L)81k19 chromosome is put in trans to zip.

View larger version (39K): [in this window] [in a new window] [Download PPT slide]   Figure 5. —Finer genetic mapping of the zipper SSNC region 73A3; 74F. Df(3L)81k19 is a strong SSNC of zip (red box). To more finely map the regions responsible for this genetic interaction, additional deficiencies were tested for SSNC of zip. Df(3L)st-b11, Df(3L)st7 and Df(3L)st4 behave as weak SSNCs (yellow box), but Df(3L)st-f13 does not interact genetically with zip (white box; Table 7). This maps a weakly interacting region to interval 73A3-4; 73C1. Based on these observations, region 74A3; 74F may also include a locus or loci that behave as SSNC of zip.

Intermediate, second-site noncomplementing third chromosome deficiencies: Two regions on the third chromosome contain loci that interact intermediately with zip.

75A6-7; 75F1: Two deficiencies, Df(3L)W10 and Df(3L)Cat, define one of these regions. They overlap between 75B8; 75C1-2. Each deficiency interacts with zip^Ebr but not with zip^6.1, and a slight increase in the penetrance of mlf is seen at 25°. The highest observed penetrance of mlf is 26% with Df(3L)W10 and 20% with Df(3L)Cat.

84D4-6; 85B6: Df(3R)p712 defines a unique interval, 84D4-6; 85B6. In all but one other case, zip^Ebr proves to be the more sensitive allele in these screens. Df(3R)p712, however, interacts with zip^6.1 but not with zip^Ebr. There is a slight temperature sensitivity observed, so at 25°, 40% of the double heterozygous flies are mlf as compared to 23% at 18°. Df(3R)Hu, Df(3R)Ant17 and Df(3R)p-Xt103 (8% mlf) are noninteracting deficiencies that overlap the proximal and distal breakpoints of Df(3R)p712. Thus the SSNC locus is confined to 84F1; 85A2.

Weak, second-site noncomplementing third chromosome deficiencies: Six regions on the third chromosome possess weak SSNC loci of zip.

62F; 63D: One region is defined at 63D1. Three overlapping deficiencies, Df(3L)M21, Df(3L)HR370 and Df(3L)HR119 overlap at 63D1 and each interacts with zip^Ebr with a penetrance between 11–22%.

63F4-7; 64C13-15: Neighboring this interval, Df(3L)GN24 also shows mlf with zip^Ebr with a penetrance of 14% at 18°. Two noninteracting deficiencies, Df(3L)GN50 and Df(3L)ZN47 (5% mlf ), overlap this region, so the interacting locus is likely to be confined to 64C.

67A2; 67D7-13: Df(3L)AC1 uncovers 67A2; 67D7-13 and is a weak SSNC of zip^Ebr at 18° and 25° (23% and 13% mlf, respectively). Comparing the cytology of this deficiency with the noninteracting deficiency Df(3L)29A6 (7%), the locus lies within 67B1; 67D7-13.

77A1; 77D1 and 78A2; 78C9: Distally on the chromosome arm 3L, nonoverlapping deficiencies, Df(3L)rdgC-co2 (11% mlf ) and Df(3L)Pc-Mk (17% mlf ), interact weakly with zip^Ebr. Df(3L)ri-79C is noninteracting and overlaps each of these deficiencies. Taken together, Df(3L)rdgC-co2 uncovers a locus in interval 77A1; 77B-C that interacts with zip^Ebr at 18° with a penetrance of mlf of 11%. Screening with Df(3L)Pc-Mk reveals a weakly interacting locus mapping to 78A; 78C9.

88E7-13; 89A1: Df(3R)Ea behaves as a weak SSNC of zip^Ebr. Testing at 25° shows that Df(3R)Ea interacts specifically with zip^Ebr. The penetrance of mlf in these flies is 23%. The genetic interactions observed with this deficiency are consistent with results obtained when mutations in the cytoplasmic tropomyosin (cTm) locus are tested (Table 8). Df(3R)Ea uncovers cTm, and maternal effect and lethal mutations specific for cTm have been identified (ERDELYI et al. 1995 ; TETZLAFF et al. 1996 ). A maternal effect allele of cTm, cTm^eg9, was generated by P-element insertion and a lethal allele, cTm^er4, was generated by imprecise excision of the P-element (ERDELYI et al. 1995 ). Both cTm alleles behave as SSNCs with zip^Ebr but not with zip^6.1 (Table 8; data not shown). The penetrance of mlf with each allele and zip^Ebr is similar to each other and to Df(3R)Ea (26–34% mlf ). Thus, Df(3R)Ea defines at least one locus that acts as a SSNC of zip^Ebr, and cTm appears responsible for the interaction.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Second-site noncomplementation of zipper :
Our screens for second-site noncomplementation of zipper with deficiencies from the Bloomington Stock Center Deficiency Kits assay for the malformed phenotype. These screens reveal 47 deficiencies on the X, second and third chromosomes that act as SSNCs of zip, and 111 that do not. The relative strength of the genetic interactions has been assigned based on the penetrance of mlf in zip; Df double heterozygous flies. Two of the 47 interacting deficiencies, Df(2R)Jp8 and Df(3L)81k19, are classified as strong SSNCs, with an observed penetrance of mlf in excess of 75%. Intermediate strength penetrance (25–75% mlf ) is seen with 17 of the deficiency stocks, and weak interactions (penetrance of 10–24%) are seen with 28 of the stocks. The two strong interacting deficiencies also affect the viability of flies that are double heterozygous with zip^Ebr. Expressivity of mlf is not always correlated with the degree of penetrance. Temperature sensitivity was assessed for 43 of the SSNC deficiency stocks. Thirteen of the stocks showed no significant temperature sensitivity while eight deficiencies exhibited cold sensitivity at 18°, and 22 exhibited temperature sensitivity at 25°. Generally, the zip^Ebr allele is more sensitive in these screens, such that SSNC is seen only with zip^Ebr, or SSNC is observed with both zip alleles, but with a lower penetrance when in trans to zip^6.1. The only exceptions to this are seen in tests with Df(1)G4e[L]h24i[R] and Df(3R)p712. In each of these cases, SSNC is observed only when double heterozygous with zip^6.1. Thirteen of the deficiencies interact with both zip^Ebr and zip^6.1, and they include the two strong SSNCs, five of the intermediate SSNCs and six weak SSNCs. Among this collection of 13 stocks, five overlap and define two regions. Df(2R)M41A4 and In(2R)bw[VDe2L]Cy[R] overlap in region 41A, and Df(2R)or-BR6, Df(2R)bw[VDe2L]Px[Kr] and In(2LR)Px[4] overlap in cytogenetic interval 60B-D. Thus, the SSNC deficiencies that interact with both zip^Ebr and zip^6.1 define 10 regions of the genome that are required for zip function and represent strong candidates for pursuit in the future. In addition, the 12 deficiencies that act as intermediate SSNCs, 11 of which interact with zip^Ebr and one that interacts with zip^6.1, are also strong candidates for future analysis.

Second-site noncomplementing loci generally map to the deficiencies:
Operationally, it is possible that the SSNC seen between the deficiency stocks and zip could map to a region of the chromosome other than that area that is uncovered by the deficiency. In all cases for which appropriate data are currently available, this is not the case. In these instances multiple stocks are available for screening. In the strongest case arguing against the interacting loci mapping outside the deficiency, overlapping deficiency stocks derived from different parental lines show a genetic interaction with zip, behaving with similar characteristics. Eight chromosomal regions are defined by overlapping SSNC deficiencies derived from different parental chromosomes (Table 9). In cases where overlapping deficiencies have not been tested, the identification of individual SSNC mutations that are uncovered by the deficiency provide alternate evidence that the SSNC is due to loci uncovered by the deficiency. In this study, we identified three mutations, E3.10/J3.8, vkg, and cTm, that are uncovered by SSNC deficiencies and are themselves, SSNCs of zip.

Genetic background affects the penetrance of mlf. When zip^Ebr/SM5 is outcrossed to second chromosome deficiency stocks, mlf is observed in an average of 3% of the zip^Ebr/CyO progeny. When outcrossed to Oregon-R wild-type flies, mlf is seen in 6% of the zip^Ebr/+ progeny. We expect the observed variation in the penetrance arises from genetic differences that could be present on any of the chromosomes within the stocks. The influence of genetic background is of greatest concern for those deficiencies that interact with zip with weak penetrance (10–24% mlf ), particularly those with an observed penetrance of 10–15%. Indeed, our selection of 10% penetrance as the minimum point for defining interaction is somewhat arbitrary, but we believe it is valid. Genetic background is unlikely to be responsible for the observed interaction with all of the weakly penetrant deficiencies. First, of the 28 identified weak class deficiencies, only 11 interact with a penetrance of 10–15%. The influence of genetic background on the interaction has been tested, albeit indirectly, for five of these 11 deficiencies. For example, deficiencies Df(2L)TE29Aa-11, Df(2L)J39 and Df(2R)or-BR6 interact with both zip^Ebr and zip^6.1 (Table 2). Because each of these zip alleles was derived independently from different parental stocks (GOTWALS and FRISTROM 1991 ; YOUNG et al. 1993 ) the genetic background for these interacting deficiencies was varied in the double heterozygotes, and yet each deficiency is interacting. Further, Df(2R)vg-C and Df(2R)CX1 are weakly interacting, overlapping SSNC deficiencies derived from different parental stocks that behave similarly in this assay (Table 9). Thus, genetic background is unlikely to solely account for the observed interactions between these deficiencies and zip.

Overall, the general applicability of this genetic screen holds strong. To date, based on the criteria listed above, of the 47 SSNC deficiencies, 11 SSNC genomic regions have been definitively mapped, and one additional interval is likely to be specific for SSNC with zip. Clearly, verification that SSNC maps to the deficiency and not elsewhere in the genome is a first step in pursuing the identification of individual SSNC loci.

Molecular basis of second-site noncomplementing loci:
The expectation of this screen is that loci that are required for myosin function during morphogenesis will be identified. Morphogenetic events that are driven by myosin-based contraction require the appropriate temporal and spatial function of myosin. Players in this process may include the following: cell surface receptors and their ligands that respond to developmental cues; intercellular signaling molecules; transcription factors that direct the synthesis of factors required upstream of myosin function; structural factors, such as other cytoskeletal proteins and extracellular matrix proteins, involved in the appropriate targeting and maintenance of myosin within the cell; and direct, myosin regulatory proteins, such as myosin light chain kinase, that regulate the activity of myosin-based contractions. In this study, the mlf leg phenotype provides the assay for this screen. Therefore, the genomic regions identified in these screens must include loci that are required to collaborate with myosin during imaginal leg disc elongation. It is expected that some of the identified loci will be specific for adult leg morphogenesis, but other loci will encode gene products that are used throughout all stages of development for cell shape change driven morphogenesis. To date, our analysis of loci uncovered by SSNC deficiencies demonstrates that a gene encoding a cytoskeletal protein and one encoding an extracellular matrix protein are each required in concert with myosin for proper imaginal leg disc morphogenesis. In addition, analysis of interactions between zip and E74 reveals that the function of myosin during leg morphogenesis may not be under general transcriptional control of ecdysone-induced transcription factors. Finally, a currently uncharacterized EMS-induced mutation that behaves as a strong SSNC of zip has been identified.

Cytoskeletal and extracellular matrix loci:
This screen implicates a function for cytoplasmic tropomyosin during morphogenesis. Drosophila cytoplasmic tropomyosin (cTm) is one product of the complex tropomyosin locus. Differential splicing and differential poly-adenylation gives rise to muscle and nonmuscle(cytoplasmic) specific transcripts and protein products (HANKE and STORTI 1988 ). P-element insertional mutations in the cTm gene were identified either as female-sterile mutations that disrupt oskar RNA localization (ERDELYI et al. 1995 ) or as zygotic lethal mutations that affect embryonic head morphogenesis (TETZLAFF et al. 1996 ). Germline-specific mutant alleles reduce the prevalence of some of the cTm-specific transcripts in the ovary (ERDELYI et al. 1995 ). Presumably, the lethal alleles would have a more severe effect on the expression of cTm, although this was not shown. A germline-specific mutation, cTm^eg9, and a lethal allele generated by P-element excision, cTm^er4, behave as SSNCs of zip during metamorphosis. This reflects a postembryonic requirement for the cTm, and that, in our assay, the "germline specific" allele has postoogenic defects. The nature of the requirement for cTm in myosin function during this morphogenetic event is presently unclear. Cytoplasmic tropomyosin is thought to stabilize actin filaments (reviewed in PITTENGER et al. 1994 ). Genetic analysis in yeast reveals this tropomyosin function. Mutations in the Saccharomyces cerevisiae tropomyosin gene lead to the loss of actin cables and a reduced growth rate (LIU and BRETSCHER 1989 ). In Schizosaccharomyces pombe, tropomyosin is localized to the contractile ring during cytokinesis, and mutations in the gene disrupt cytokinesis (BALASUBRAMANIAN et al. 1992 ). Based on these observations, cTm mutations are likely to disrupt myosin function during leg elongation by disrupting the actin cytoskeleton.

Collagen IV is a basement membrane collagen, and its function has been demonstrated during morphogenesis in C. elegans and Drosophila. Mutations in the emb-9 locus [Collagen 1(IV)] and the let-2 locus [Collagen 2(IV)] cause defects during the late morphogenetic stage that result in embryonic lethality (GUO et al. 1991 ; SIBLEY et al. 1993 ). In Drosophila, two Collagen IV genes, DCg1 [1(IV); NATZLE et al. 1983 ] and viking [2(IV); YASOTHORNSRIKUL et al. 1997 ], localize to the SSNC interval uncovered by Df(2L)sc19-5. Mutations in viking act as SSNCs of zip during leg morphogenesis. Levels of Collagen IV are detectable throughout the life cycle of the fly, with high levels detected until the onset of pupation (FESSLER et al. 1993 ). During leg morphogenesis, the basal lamina detaches from the disc epithelium, and proteolysis of Collagen IV is thought to make the basement membrane more extensible, thus facilitating leg elongation (FESSLER et al. 1993 ). This site-directed cleavage of Collagen IV occurs in response to ecdysone (BIRR et al. 1990 ; FESSLER et al. 1993 ). Identification of Stubble as a necessary gene product during leg imaginal disc morphogenesis also demonstrates a role for proteolysis during this process. The Stubble locus encodes a type II transmembrane serine protease (APPEL et al. 1993 ), and it interacts genetically with broad¹ and zip^Ebr (GOTWALS and FRISTROM 1991 ). However, since the existing viking mutations are P-element insertions, it is more likely that these mutations affect the expression of Collagen IV, rather than its proteolysis, during metamorphosis. Basal cytoplasmic extensions have been observed in the imaginal discs of several insects (reviewed in FRISTROM and FRISTROM 1993 ) and are thought to be important for morphogenetic movements. Perhaps mutations in viking interfere with such processes in the Drosophila leg imaginal disc, and this effect would occur prior to the observed proteolysis. Mutations are available in the DCg1 1(IV) gene (RODRIGUEZ et al. 1996 ), and it is of interest to know if these mutations also act as SSNCs of zip.

Transcription factors:
The requirement for myosin function during leg morphogenesis is demonstrated by several lines of evidence. zip^Ebr was first identified as E(br) in screens that identified genetic interactors with the ecdysone-inducible Broad-Complex (BR-C ) family of transcription factors (GOTWALS and FRISTROM 1991 ). Later, rescue with a cosmid bearing a zip transgene ( YOUNG et al. 1993 ) showed that E(br) is an allele of zip (GOTWALS 1992 ). Further, myosin is subcellularly localized within leg imaginal discs undergoing cell shape change (VON KALM et al. 1995 ). Finally, experimental depletion of myosin during larval development gives rise to mlf in adults (EDWARDS and KIEHART 1996 ). Transcription of zip is not under ecdysone control in the imaginal discs (VON KALM et al. 1995 ); therefore, the gene expression directed by BR-C must affect other aspects of leg disc morphogenesis, rather than merely inducing zip expression. E74 also encodes ecdysone-inducible transcription factors whose expression is induced in late third instar larvae (BURTIS et al. 1990 ). E74 is a complex locus, encoding two transcripts, each of which has been mutated separately (BURTIS et al. 1990 ; FLETCHER et al. 1995 ). Genetically, BR-C and E74 mutations interact, suggesting they function in overlapping pathways (FLETCHER and THUMMEL 1995 ). Pupae hemizygous for the E74^DL-1 mutation exhibit, among other phenotypes, a shortened leg phenotype (FLETCHER et al. 1995 ). Df(3L)81k19 uncovers the E74 locus, and it is a strong SSNC of zip. However, mutations in each of the E74 transcripts, either singly or in trans to each other, do not interact with zip. Clearly a locus other than E74 on the Df(3L)81k19 chromosome is responsible for the SSNC. Finer genetic mapping reveals a more weakly interacting locus may reside in cytogenetic region 73A3-4; 73C1. Identification of the other SSNC locus (loci) uncovered by the Df(3L)81k19 chromosome awaits finer genetic analysis.

Conclusion:
The goal of these screens is to identify factors, both known and novel, that affect myosin function during morphogenesis. In fact, both known and novel loci have been identified. Characterization of SSNC deficiencies identified in this study indicates that they uncover an expected array of loci, including those that encode cytoskeletal and extracellular matrix proteins and a transcription factor, that are required for myosin function during leg disc morphogenesis. While the role of the Broad-Complex and perhaps viking may be confined to imaginal disc morphogenesis, it is clear that the functional interactions between myosin and cytoplasmic tropomyosin are likely to occur in morphogenetic and cell migration events during oogenesis, embryogenesis and metamorphosis. In addition, these screens reveal that a previously uncharacterized mutation, E3.10/J3.8 is a strong interactor with zip, and that it is required for both morphogenesis and viability in the fly. Phenotypic and molecular characterization of the E3.10/J3.8 mutation will proceed along with further characterization of the interacting deficiencies identified in these screens. These studies promise to extend our understanding of the spatial and temporal contr