Morphogenesis is characterized by orchestrated changes in the shape and position of individual cells. Many of these movements are thought to be powered by motor proteins. However, in metazoans, it is often difficult to match specific motors with the movements they drive. The nonmuscle myosin II heavy chain (MHC) encoded by zipper is required for cell sheet movements in Drosophila embryos. To determine if myosin II is required for other processes, we examined the phenotypes of strong and weak larval lethal mutations in spaghetti squash (sqh), which encodes the nonmuscle myosin II regulatory light chain (RLC). sqh mutants can be rescued to adulthood by daily induction of a sqh cDNA transgene driven by the hsp70 promoter. By transiently ceasing induction of the cDNA, we depleted RLC at specific times during development. When RLC is transiently depleted in larvae, the resulting adult phenotypes demonstrate that RLC is required in a stage-specific fashion for proper development of eye and leg imaginal discs. When RLC is depleted in adult females, oogenesis is reversibly disrupted. Without RLC induction, developing egg chambers display a succession of phenotypes that demonstrate roles for myosin II in morphogenesis of the interfollicular stalks, three morphologically and mechanistically distinct types of follicle cell migration, and completion of nurse cell cytoplasm transport (dumping). Finally, we show that in sqh mutant tissues, MHC is abnormally localized in punctate structures that do not contain appreciable amounts of filamentous actin or the myosin tail-binding protein p127. This suggests that sqh mutant phenotypes are chiefly caused by sequestration of myosin into inactive aggregates. These results show that myosin II is responsible for a surprisingly diverse array of cell shape changes throughout development.

The pathway from a developmental decision to a morphological change in a cell or tissue can typically be divided into two sets of events. The first is a cascade of regulatory processes, such as cell-cell signals and changes in gene expression. Many of the molecules involved in these processes can be readily identified by molecular or genetic screens, because their expression patterns or mutant phenotypes are tissue-specific. However, these molecules have no direct impact on cell morphology. The second set of events includes activation of ‘effector’ molecules that actually drive cell shape changes. Effectors are often cell adhesion and cytoskeletal proteins that tend to be generally expressed and have multiple and/or redundant functions. These properties can make the genetic analysis of effectors rather difficult; as a result, effectors of many developmental processes remain unidentified or poorly characterized.

Drosophila provides not one, but several outstanding model systems to study the regulation and mechanics of tissue morphogenesis. Adult eyes, wings, ovaries and testes can yield a wide range of mutant phenotypes because each of these organs is highly differentiated, but dispensable for viability of the fly. Many upstream regulators of morphogenesis have been characterized for these and other structures. The connection between these regulators and their effectors may be elucidated by a variety of methods including genetic interaction strategies and ‘reverse’ genetics (disrupting the function of proteins suspected to be important).

We have performed a reverse genetic analysis of the actinbased motor protein nonmuscle myosin II. Numerous studies have shown that the actin cytoskeleton plays a major role in tissue morphogenesis and can produce specific forces within the cell in direct response to external signals (aspects of this are reviewed by von Kalm et al., 1995; Hall, 1994). Of all the actin modulating proteins that can influence cell shape, the myosins are uniquely suited to generate contractile forces by sliding actin filaments past each other or along the plasma membrane. However, the in vivo composition, organization and regulation of an actomyosin contractile system are well described only in the specialized case of muscle cells, where contraction is powered by sarcomeric myosin II (Harrington and Rodgers, 1984; Rayment et al., 1993). In nonmuscle cells, myosin II is ubiquitous and has been suspected to drive numerous motile events (Warrick and Spudich, 1987). However, all eucaryotes studied express multiple, distinct types of myosin, as well as numerous other motor proteins, complicating the assignment of specific roles to each motor (Cheney et al., 1993). The clearest example of a nonmuscle myosin-driven shape change is animal and amoeboid cytokinesis, in which an actin and myosin-rich contractile ring cleaves the cell. Numerous studies demonstrate that nonmuscle myosin II (henceforth, ‘myosin’) is required for contractile ring function (Mabuchi and Okuno, 1977; DeLozanne and Spudich, 1987; Karess et al., 1991; Chen et al., 1994). Myosin is also essential for development in Dictyostelium (DeLozanne and Spudich, 1987; see Discussion) and metazoans (Young et al., 1993; Wheatley et al., 1995).

Among metazoans, the only genetic studies of nonmuscle myosin II function have been performed in Drosophila. Drosophila myosin was purified and the single-copy genes encoding its three subunits, myosin heavy chain (MHC), regulatory light chain (RLC) and essential light chain (ELC), were cloned and sequenced (Kiehart and Feghali, 1986; Kiehart et al., 1989; Ketchum et al., 1990; Karess et al., 1991; Edwards et al., 1995; Mansfield et al., 1996). Null alleles of zipper, which encodes MHC, cause failure of dorsal closure, during which the lateral epidermal cells elongate to cover the dorsal surface of the embryo after germ band retraction. Axon pathfinding and head involution defects also contribute to the embryonic lethality of zipper nulls (Young et al., 1993). A hypomorphic allele of spaghetti squash (sqh), encoding RLC, leads to extensive failure of cytokinesis in larvae (Karess et al., 1991). Together these phenotypes show that the same myosin is required for both cell division and the morphogenesis of epithelial sheets. These mutant phenotypes do not reflect all myosin functions, only those functions that are visibly disrupted when both maternal and zygotic myosin become insufficient. Therefore, by inactivating myosin at different developmental times one could uncover additional functions; for example, early embryos injected with myosin antibodies or derived from sqh germline clones show defects in the arrangement of the syncytial nuclei, suggesting a role for myosin in nuclear migration (Kiehart et al., 1990; Wheatley et al., 1995).

Here we present several newly identified, unexpected developmental functions of myosin by rescuing sqh mutations in a reversible and temporally controlled manner. We confirm that, in sqh mutant cells, MHC forms nonfunctional aggregates, suggesting that sqh phenotypes are due to a simple reduction in myosin activity. With daily induction, a heat-shockpromoter-driven sqh cDNA rescues both the original weak sqh1 allele and a newly generated severe allele, sqh2. By withholding induction of the sqh cDNA, myosin function can be greatly reduced at particular developmental stages. This approach reveals that myosin is required for several distinct forms of cell movement during morphogenesis of imaginal discs and egg chambers.

Heat-shock-dependent rescue of sqh

An hsp70 promoter-driven sqh cDNA (P[w+, hs-sqh+], referred to as hs-sqh+) was produced by ligating a 0.6 kb EcoRI fragment from a partial sqh cDNA in λgt11 and the EcoRI-NotI fragment of the fulllength sqh cDNA in pNB40, into EcoRI-NotI cut pCaSpeR-hs P element vector (Karess et al., 1991; Thummel and Pirotta, 1992; Lindsley and Zimm, 1992). Multiple transgenic lines bearing different insertions of this construct were obtained by injecting the construct into w; Δ2-3 embryos by standard methods (Robertson et al., 1988; Ashburner, 1989). Several lines gave indistinguishable results. Heatshock induction of the transgene was accomplished by placing glass culture vials (9.5×2.5 cm, weight 28 g including medium) in a 37°C room for 50-60 minutes. For complete rescue of sqh, we used a standard regimen of one heat shock per day (termed ‘HS’). For partial rescues, a given number of daily heat shocks was omitted. Apart from heat shock, cultures were kept at 21-22°C (or 25°C for Fig. 1A).

Fig. 1.

Phenotypes due to depletion of RLC at various developmental stages. (A) Alternate 1-2 day broods from sqh1; hs-sqh+ flies were raised with (+) or without (−) daily heat shock. Without heat shock, most larvae fail to evert their anterior spiracles and die at pupariation with no sign of imaginal development (black boxes). Some evert their anterior spiracles but develop no adult cuticular structures (dark gray boxes). With daily heat shock, genetically identical larvae typically develop into normal adults (white boxes) or form adult structures but die in the pupal case (light gray boxes). Since alternate broods show consistent phenotypes, only heat-shock induction of hs-sqh+ (not other environmental or maternal factors) is responsible for rescue of sqh1. (B-D) A series of 1 day broods from sqh1; hs-sqh+ flies was heat shocked daily except for one 72 hour gap between heat shocks. Terminal phenotypes of each brood are plotted versus the age of the brood at the end of the 72 hour gap (‘age at the point of lowest RLC accumulation’). ‘No HS’ and ‘+’ columns were raised with no heat shocks, or no gap, respectively. Eye phenotypes were categorized by estimating the % of ommatidia that deviated from the normal hexagonal packing. (C) Pupal cases were scored for % with pharate adults. (D) % of adults with bristle defects, and the ‘malformed’ syndrome of leg and wing defects, were independently scored. Broods lose synchrony as they develop, but the typical stage at each age is noted above (B). *, approximate time of eye disc morphogenetic furrow initiation (Wolff and Ready, 1993).

Fig. 1.

Phenotypes due to depletion of RLC at various developmental stages. (A) Alternate 1-2 day broods from sqh1; hs-sqh+ flies were raised with (+) or without (−) daily heat shock. Without heat shock, most larvae fail to evert their anterior spiracles and die at pupariation with no sign of imaginal development (black boxes). Some evert their anterior spiracles but develop no adult cuticular structures (dark gray boxes). With daily heat shock, genetically identical larvae typically develop into normal adults (white boxes) or form adult structures but die in the pupal case (light gray boxes). Since alternate broods show consistent phenotypes, only heat-shock induction of hs-sqh+ (not other environmental or maternal factors) is responsible for rescue of sqh1. (B-D) A series of 1 day broods from sqh1; hs-sqh+ flies was heat shocked daily except for one 72 hour gap between heat shocks. Terminal phenotypes of each brood are plotted versus the age of the brood at the end of the 72 hour gap (‘age at the point of lowest RLC accumulation’). ‘No HS’ and ‘+’ columns were raised with no heat shocks, or no gap, respectively. Eye phenotypes were categorized by estimating the % of ommatidia that deviated from the normal hexagonal packing. (C) Pupal cases were scored for % with pharate adults. (D) % of adults with bristle defects, and the ‘malformed’ syndrome of leg and wing defects, were independently scored. Broods lose synchrony as they develop, but the typical stage at each age is noted above (B). *, approximate time of eye disc morphogenetic furrow initiation (Wolff and Ready, 1993).

Generation of sqh2

The original allele sqh1, recovered from a hybrid dysgenic cross, has an unmarked 0.7 kb P element inserted in the 5′ untranslated region of exon 2 (Karess et al., 1991). To obtain an imprecise excision of this P element, we removed most of the strain’s extraneous P elements by outcrossing and recombination, generated males of the genotype sqh1/Y; P[w+, hs-sqh+]; Δ2-3 (transposase gene; Robertson et al., 1988) and screened the progeny of these males by PCR for a deletion between two primer binding sites that flank the sqh P element. A new allele, sqh2, was recovered and characterized by PCR mapping and direct sequencing of PCR products. sqh2 is missing most of the P element and 1.4 kb upstream of it, but is rescued by P[w+, sqh+] and therefore does not lack any other vital DNA upstream of sqh. Since it retains the sqh coding region, it may produce some intact RLC if the deletion moved a cryptic promoter in front of exon 2. However, we classify sqh2 as a null or near null based on the following observations. First, the homozygous sqh2 phenotype is much more severe than that of sqh1, which makes roughly 5-10% as much mRNA as sqh+ (Karess et al., 1991; K. A. E, unpublished). Second, unlike sqh1, the sqh2 phenotype is not noticeably more severe in female hemizygotes (sqh2/Df(1)5D; Karess et al., 1991). Third, when maternal RLC contribution is reduced, sqh2 becomes embryonic lethal displaying phenotypes both identical to, and more severe than, MHC nulls (J. Crawford, K. A. E., D. P. K., unpublished observations). Finally, no sqh transcript is discernable on northern blots of RNA from sqh2 homozygotes (M. Champagne, K. A. E., D. P. K., unpublished observations).

Immunolocalization and microscopy

Anti-nonmuscle myosin heavy chain polyclonal antiserum (656) is described by Kiehart and Feghali (1986); anti-α spectrin by Pesacreta et al. (1989). Anti-p127 antibodies were kindly provided by B. Mechler (Strand et al., 1994a,b). Ovaries dissected in EBR (Montell et al., 1992) were fixed in fresh 2% paraformaldehyde in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) for 10 minutes, washed with PBS, blocked in PBT (=PBS with 0.2% Triton X-100) plus 3% goat serum for 30 minutes, incubated with 656 antiserum (1:2000 in PBT with 1% goat serum), washed and stained with Cy5 or Cy3 conjugated anti-rabbit secondary antibody (Jackson ImmunoResearch, West Grove, PA). Incubations were performed overnight at 4°C or 4 hours at room temperature. To label filamentous (F-) actin, samples were stained with rhodamine-labelled phalloidin (Molecular Probes, Eugene, OR) for 20 minutes before final washes (all washes were 4× with PBT with 1% goat serum over 40 minutes). Samples were mounted in SPIF antifade/ DNA stain (Lundell and Hirsh, 1994) and viewed with a Zeiss LSM 410 confocal microscope. For double or triple labelling, the following excitation/emission filter sets (nm) were used: Cy3 or rhodamine: 568/ >590; Cy5: 647/ 670-810; SPIF DNA stain: 488/ 515-540 or 670-810. Using Adobe Photoshop software, images were overlaid and, for SPIF and some phalloidin images, contrast was uniformly enhanced using ‘Curves’ to suppress bleedthrough from other channels. Original contrast is shown for all antibody-stain images and the phalloidin images in Figs 3 and 8. Final display colors were chosen for clarity rather than to mimic the original fluorochrome.

To evaluate their morphology, mature eggs were dissected from ovaries in EBR, then either fixed as above and mounted in PBS, or mounted unfixed in PBT, and digital micrographs were taken with differential interference contrast optics. Adult structures were stored in 70% ethanol, mounted in CMCP10 (Masters Chemical Co., Elk Grove, IL) and viewed with transmitted light. Electron microscopy specimens were dehydrated in Peldri II (Ted Pella, Inc., Redding CA), coated and examined with a Phillips 501 SEM.

sqh2 is an early larval lethal mutation

We generated a severe sqh allele lacking the 5′ noncoding portion of the gene (sqh2, see Materials and Methods) because the original sqh1 allele produces small but significant amounts of RLC (Karess et al., 1991). Nearly all sqh2 embryos hatch and molt to 3rd instar, but fail to grow in size, remaining the length of 2nd instar larvae. They invariably become sluggish and die 4-8 days after hatching (all times given in this paper refer to development at 21-22°C). In contrast, sqh1 larvae grow to full size, experience a prolonged larval period of several weeks and die just after pupariation. For comparison, wild-type and heterozygous sqh larvae pupariate at about 9-10 days after hatching. No morphological defects are apparent in sqh2 larvae except that the imaginal tissues are rudimentary and highly polyploid, as in sqh1 (data not shown). A P element transgene bearing a 3.0 kb genomic DNA fragment containing the RLC promoter and transcription unit (P[w+, sqh+], Karess et al., 1991) fully rescues both sqh1and sqh2.

sqh is rescued by an inducible RLC gene

Both sqh alleles suppress cytokinesis, precluding formation of imaginal discs. Thus, these mutations alone cannot be used to study the role of myosin in disc morphogenesis or subsequent development of adult structures. To circumvent this problem, we fused a sqh cDNA to the heat-inducible hsp70 promoter and produced transformants bearing this construct (designated hs-sqh+) on the second chromosome. hs-sqh+ fully rescues all identifiable defects in sqh animals, but only when they are subjected to a daily heat-shock regimen (termed HS) during development (see Materials and Methods). Even the severe sqh2 allele can be fully rescued by induction of hs-sqh+. These rescued flies probably lack all endogenous sqh transcriptional regulatory elements, indicating that precise transcriptional control of sqh is unnecessary for survival as long as a uniform, daily supply of sqh transcript is provided. Flies rescued by hs-sqh+ are fertile, allowing us to maintain homozygous sqh mutant stocks for indefinite periods of time. However, when removed from the daily heat-shock regimen (‘without HS’), these stocks quickly develop mutant phenotypes (see below) and die out.

To better characterize HS-dependent rescue, sqh1; hssqh+ males and females were mated and their eggs were collected in a series of vials. Alternate vials were either kept at 25°C steadily or heat shocked daily (‘on HS’), and the terminal phenotypes of all visible larvae, pupae and adults were scored (Fig. 1A). Without HS, less than 0.5% of larvae (4 of 867) made any adult cuticular structures and none eclosed. The major lethal period without HS was identical to that of unrescued sqh1 larvae (Karess et al., 1991). Their siblings on HS most often eclosed as normal adults, though a portion died as pharate adults, perhaps due to the stress of HS or a reduction in myosin function during a brief critical period. Similar numbers of animals are obtained under both conditions (though these numbers decrease in later broods as the mothers age and lay fewer eggs). These results show that without HS, the transgene provides almost no RLC activity, but with HS, it provides enough activity to rescue sqh fully and with high frequency.

Partial rescue of sqh to dissect temporal requirements for RLC

Novel phenotypes appear in sqh1; hs-sqh+ animals when daily induction of RLC is halted only temporarily. When one daily heat shock is omitted, resulting in a 48 hour period without induction, sqh1; hs-sqh+ larvae develop normally. However, omitting two heat shocks (72 hours without induction) leads to several defects. The simplest interpretation of our results is as follows. Each heat-shock induction provides a burst of new RLC transcript and protein, which are subsequently degraded at some rate. 48 hours after an induction, enough RLC remains to satisfy all requirements for myosin. By 72 hours, RLC concentration falls below a critical level, disrupting a process that requires large quantities of myosin at that time. Thus, by allowing RLC to be depleted at specific times and scoring the resulting phenotypes, we can determine when myosin is strongly required for various developmental processes. The results of such an experiment are shown in Fig. 1B-D. Consecutive one-day collections of eggs from sqh1; hs-sqh+ parents were maintained on the daily heat-shock regimen except that two consecutive heat shocks were omitted, resulting in one 72-hour gap between heat shocks. The phenotypes of all resulting adults were scored and fell into four major categories: eclosed with eyes having disrupted ommatidia (Fig. 1B), failed to eclose (Fig. 1C), eclosed with the ‘malformed’ syndrome of leg and wing defects, and eclosed with one or more doubled or missing (but morphologically normal) scutellar bristles (Fig. 1D). The frequencies of the eye and leg/wing defects are strongly correlated with the age at which the larvae were removed from HS. In contrast, the bristle defects were not tied to a specific developmental period and so might represent a general requirement for RLC or sensitivity of flies to the effects of heat shock. A similar experiment using the severe allele sqh2 also yields these phenotypes, but they appear after a shorter gap (48 versus 72 hours) between heat shocks. Presumably less RLC accumulates in sqh2 than in sqh1 larvae, so that RLC concentration more quickly falls below the critical level.

RLC is strongly required for eye morphogenesis

The sqh eye phenotype ranges in severity from slight defects to complete disorder in the usually perfect hexagonal packing of the ommatidia and severe cases usually include an anterior notch in the eye similar to the Bar phenotype (Fig. 2B). Both the frequency and severity of the eye phenotype peaked strongly when hs-sqh+ was not induced on days 5-6 after hatching. In well-synchronized cultures, this heatshock regimen results in 100% penetrance of the eye phenotype. For both sqh alleles, withholding heat shock earlier or later in development results in few or no eye defects. sqh/+; hs-sqh+ flies never show disrupted ommatidia under any of the heat-shock regimens that we used. These results demonstrate that the eye disc requires RLC most acutely at day 6-7 after hatching (about 2 days prior to pupariation), the time at which the morphogenetic furrow is initiated in the eye disc (Wolff and Ready, 1993).

Fig. 2.

RLC depletion disrupts imaginal disc morphogenesis. (A) SEM of wild-type adult eye. (B) Eye phenotype that results when sqh1; hs-sqh+ larvae are removed from HS for 72 hours such that RLC accumulation is lowest at day 5-6 after hatching, about 4 days prior to pupariation. The ommatidia are not aligned, some are fused and the eye has an anterior notch. Anterior is left in A,B. (C-E) Leg disc eversion is compromised in sqh pupae. (C) Typical rear leg from a sqh2; hs-sqh+ adult, raised on HS, is identical to wild type (not shown). (D,E) Both rear legs from a sqh2; hs-sqh+ adult, raised on HS except for a 48 hour period ending 6 days before eclosion (or approximately 10 days after hatching). Legs are bent, shortened and thickened, characteristic of the ‘malformed’ syndrome. Bars, 100 μm.

Fig. 2.

RLC depletion disrupts imaginal disc morphogenesis. (A) SEM of wild-type adult eye. (B) Eye phenotype that results when sqh1; hs-sqh+ larvae are removed from HS for 72 hours such that RLC accumulation is lowest at day 5-6 after hatching, about 4 days prior to pupariation. The ommatidia are not aligned, some are fused and the eye has an anterior notch. Anterior is left in A,B. (C-E) Leg disc eversion is compromised in sqh pupae. (C) Typical rear leg from a sqh2; hs-sqh+ adult, raised on HS, is identical to wild type (not shown). (D,E) Both rear legs from a sqh2; hs-sqh+ adult, raised on HS except for a 48 hour period ending 6 days before eclosion (or approximately 10 days after hatching). Legs are bent, shortened and thickened, characteristic of the ‘malformed’ syndrome. Bars, 100 μm.

RLC is required during leg disc eversion

Depletion of RLC at pupation often results in a collection of defects termed the malformed syndrome (mlf). Mlf includes curved, thickened upper legs (Fig. 2D,E) and defective wings, and was first observed in flies carrying mutations in the BroadComplex and any one of several loci identified as enhancers of broad (reviewed by von Kalm et al., 1995; Fristrom and Fristrom, 1993). In sqh flies with severe cases of mlf, the wings are usually folded in half, with the fold parallel to the crossveins and occasionally the tip of the wing is attached to the lateral thorax just anterior to the haltere (not shown). Mlf is most penetrant when myosin function is at a minimum about 10 days after hatching, when most animals are undergoing eversion of the leg imaginal discs (Fig. 1D). Flies that avoid mlf under these conditions usually do so in two ways: they frequently die before eclosion (Fig. 1C), or else delay pupal development until they are returned to HS. This result suggests that the cell shape changes that drive leg disc eversion are myosin-II dependent (see Discussion).

Adult RLC requirements

After eclosion, the low levels of RLC made by sqh1 are sufficient for survival of rescued adults. Even the severe sqh2 allele allows rescued adult males to survive for over a month without further heat-shock induction of RLC, though most sqh2 females die within 5-10 days. Surprisingly, both sqh1 and sqh2 males remain fertile without induction of RLC, though spermatogenesis may nonetheless be compromised (not tested). Rescued females, however, become completely sterile without continued RLC induction, due to a variety of defects in egg chamber morphogenesis. Even after prolonged sterility (up to three weeks for sqh1; hs-sqh+, or several days for sqh2; hs-sqh+), returning females to HS can restore fertility, demonstrating that all oogenesis phenotypes are solely the result of RLC depletion.

Because oogenesis provides an excellent model system to study the role of the cytoskeleton in development, we characterized sqh ovary phenotypes in detail. We present first myosin’s wildtype distribution, then the effects of short-term depletion of myosin and finally the more severe effects of longterm myosin deprivation. By correlating sqh phenotypes with specific wild-type myosin structures, we demonstrate several distinct cellular functions for myosin.

Myosin localization during oogenesis

Myosin was localized by indirect immunofluorescence in wild-type ovaries. Each egg chamber is a cluster of germ cells (15 nurse cells and one oocyte) packaged in a somatic epithelium, the follicle cell layer (see Spradling, 1993 for a review of oogenesis). Egg chambers form in the germarium and constantly bud from its posterior end. As they exit the germarium, egg chambers become separated by specialized follicle cells that form interfollicular stalks (Fig. 3A). In cells of the germaria and early egg chambers, anti-nonmuscle myosin staining is evenly distributed throughout the cytoplasm and concentrated at the cortex (Fig. 3A-C,F), as observed in the embryo (Young et al., 1991). As egg chambers mature, myosin staining becomes tightly apposed to the nurse cell membranes and is observed in the ring canals, the stable, actin-rich structures that line the cytoplasmic bridges that interconnect the germ cells (Fig. 3B-E, 4C,D; see Mahajan-Miklos and Cooley, 1994, for a review of ring canals). In later stages, staining is most intense at the basal (outer) ends of the follicle cells and is uniform among these cells up to stage 9 (Fig. 3F). In chambers punctured just before fixation, the basal ends are still brighter, indicating this pattern is not an artifact of poor antibody penetration (Fig. 3F).

Fig. 3.

Confocal immunofluorescence micrographs of wild-type actin and myosin distributions during oogenesis. (A) Myosin staining in the germarium (left) and early stage egg chambers; arrow indicates one of the interfollicular stalks, which stain prominently. (B,C) Double staining (B, myosin; C, actin) of an early stage 9 chamber. A ring canal (r)is visible in both panels. The border cells (b) are delaminating from the surrounding follicular epithelium. They exhibit actin- and myosin-rich apical processes (extending toward the right). (D,E) Double staining (D, myosin; E, actin) of a late stage 9 chamber. Border cell cluster shows both forward (arrow) and rearward accumulations of myosin. Centripetal cells are beginning to accumulate apical actin and myosin (see Fig. 4). Muscle sheaths (m) stain brightly with phalloidin (C, E), but not with nonmuscle myosin-specific serum (B, D); note there is no bleedthrough of actin signal into the myosin channel. (F) Myosin staining of a stage 8 chamber punctured with a tungsten needle shortly before fixation. The nurse cell above the puncture site (p) shows slightly increased cytoplasmic myosin staining, but the ratio of apical (a) to basal staining in the follicle cells is not affected by their increased accessibility to the antibody near the puncture. (G) Schematic diagram of the three follicle cell migrations of stage 9-10 chambers. Arrows indicate direction of movement. BC, border cells (dark gray); CC, centripetal cells (black); OFC, oocyte follicle cells (light gray), NC, nurse cells. For all panels, anterior is left and bars represent 20 μm.

Fig. 3.

Confocal immunofluorescence micrographs of wild-type actin and myosin distributions during oogenesis. (A) Myosin staining in the germarium (left) and early stage egg chambers; arrow indicates one of the interfollicular stalks, which stain prominently. (B,C) Double staining (B, myosin; C, actin) of an early stage 9 chamber. A ring canal (r)is visible in both panels. The border cells (b) are delaminating from the surrounding follicular epithelium. They exhibit actin- and myosin-rich apical processes (extending toward the right). (D,E) Double staining (D, myosin; E, actin) of a late stage 9 chamber. Border cell cluster shows both forward (arrow) and rearward accumulations of myosin. Centripetal cells are beginning to accumulate apical actin and myosin (see Fig. 4). Muscle sheaths (m) stain brightly with phalloidin (C, E), but not with nonmuscle myosin-specific serum (B, D); note there is no bleedthrough of actin signal into the myosin channel. (F) Myosin staining of a stage 8 chamber punctured with a tungsten needle shortly before fixation. The nurse cell above the puncture site (p) shows slightly increased cytoplasmic myosin staining, but the ratio of apical (a) to basal staining in the follicle cells is not affected by their increased accessibility to the antibody near the puncture. (G) Schematic diagram of the three follicle cell migrations of stage 9-10 chambers. Arrows indicate direction of movement. BC, border cells (dark gray); CC, centripetal cells (black); OFC, oocyte follicle cells (light gray), NC, nurse cells. For all panels, anterior is left and bars represent 20 μm.

Fig. 4.

A ring of myosin is present during centripetal cell ingression. One early stage 10 egg chamber (A,B) and one late stage 10 egg chamber (C,D), stained using anti-myosin, were optically sectioned by confocal microscopy (anterior is left). (A,C) Single sections show that the centripetal cells have an increased accumulation of myosin, some of which concentrates at their apical ends (white arrows). All other oocyte follicle cells (f) accumulate myosin only at the basal (outer) ends. Myosin is also concentrated in the border cells and ring canals, visible between the arrows in A and C, respectively. (B,D) Zseries of the same pair of egg chambers were collected and rotated (anterior ends tilted back into the page). This perspective shows that the apical myosin in the centripetal cells connects to form a continuous ring (black arrows). The ring remains but decreases in diameter as these cells move centripetally (direction of arrows). n, nurse cells; o, oocyte; bars, 25 μm.

Fig. 4.

A ring of myosin is present during centripetal cell ingression. One early stage 10 egg chamber (A,B) and one late stage 10 egg chamber (C,D), stained using anti-myosin, were optically sectioned by confocal microscopy (anterior is left). (A,C) Single sections show that the centripetal cells have an increased accumulation of myosin, some of which concentrates at their apical ends (white arrows). All other oocyte follicle cells (f) accumulate myosin only at the basal (outer) ends. Myosin is also concentrated in the border cells and ring canals, visible between the arrows in A and C, respectively. (B,D) Zseries of the same pair of egg chambers were collected and rotated (anterior ends tilted back into the page). This perspective shows that the apical myosin in the centripetal cells connects to form a continuous ring (black arrows). The ring remains but decreases in diameter as these cells move centripetally (direction of arrows). n, nurse cells; o, oocyte; bars, 25 μm.

Myosin in migrating and ingressing follicle cells

At stage 9, a series of follicle cell migrations begins (Fig. 3G). The nearly uniform follicular epithelium splits into two cell populations. Roughly 8 cells at the anterior tip of the egg chamber (border cells) delaminate and crawl between the nurse cells until they reach the oocyte (Fig 3B-E; Montell et al., 1992). While they migrate, the border cells remain tightly adherent to each other and their myosin staining is much brighter than that of the surrounding nurse cells. Border cells extend forward projections that stain brightly for myosin and actin (Fig. 3B-E).

The remaining population of follicle cells moves posteriorly over the outer surface of the nurse cells toward the growing oocyte. Those that contact the oocyte (oocyte follicle cells) become columnar, while those remaining over the nurse cells flatten and spread (Fig. 3D,E). In stage 10, the most anterior ring of oocyte follicle cells, resting on the oocyte-nurse cell border, elongate and plunge inward toward the border cells, which have come to rest at the anterior face of the oocyte (Fig. 3G). These ‘centripetal cells’ will eventually cover the anterior of the oocyte and build several specialized structures of the anterior chorion. Each centripetal cell specifically accumulates a bright bar of myosin staining at the edge of the apical (inner) surface that leads the penetration between the nurse cells and oocyte (Fig. 4). F-actin is also concentrated at this edge (Fig. 3E). These ‘leading edges’ of each centripetal cell join to form a continuous band of actomyosin staining around the egg chamber, which decreases in diameter as the cells move centripetally (Fig. 4; heavy black ring in Fig. 3G). Contraction of this actomyosin band could provide the force to pull the centripetal cells inward (see below). Myosin distribution at subsequent stages is obscured by the autofluorescent chorion, making later events such as dorsal appendage formation difficult to observe.

Short-term RLC depletion disrupts follicle cell migrations

When sqh mutant ovaries are no longer rescued by heat-shock induction of RLC, they display major defects in the migrations of three distinct follicle cell populations: the border and centripetal cells (Fig. 3G), and later the dorsal appendage cells. These defects are not due to failure of cytokinesis, because they appear even in egg chambers that had enough residual RLC to complete cytokinesis. Normally border cell migration and oocyte follicle cell migration are simultaneous at stage 9, and both are complete well before the onset of rapid nurse cell cytoplasm transport (‘dumping’, stage 11). In sqh2; hs-sqh+ females, border cell movement falls behind oocyte follicle cell migration; after 3 days without heat shock, no border cells reach their destination at the oocyte and many fail to migrate at all. Dumping is initiated, as indicated by the dramatic appearance of cytoplasmic actin cables in the nurse cells (Mahajan-Miklos and Cooley, 1994), even though frequently the border cells have not yet moved (Fig. 5). Surprisingly, as long as a reasonable number of follicle cells are present, oocyte follicle cell migration proceeds normally in spite of severe depletion of RLC.

Fig. 5.

sqh2 border and centripetal cells fail to migrate, as seen by confocal optical sectioning. Green, cell outlines revealed by rhodamine-phalloidin labelling of F-actin; red, nuclei labelled with SPIF DNA stain. (A) Wild-type egg chamber at stage 10, just prior to formation of cytoplasmic actin fibers. (B) sqh2; hs-sqh+ egg chamber, 3 days without RLC induction, which has progressed to stage 10B-11 based on the presence of cytoplasmic actin fibers (act). Follicle cells have proliferated normally. Centripetal cells (cc) elongate at stage 10 in wild-type chambers (A), but fail to elongate in sqh2 chambers (B). Wild-type border cells (bc) reach the oocyte by stage 10 (A); sqh2 border cells remain at the anterior tip of the egg chamber (B). Nuclei appear to be clogging some ring canals (B), but unlike the cell migration phenotypes, this nuclear phenotype is not seen consistently. Bar, 50 μm.

Fig. 5.

sqh2 border and centripetal cells fail to migrate, as seen by confocal optical sectioning. Green, cell outlines revealed by rhodamine-phalloidin labelling of F-actin; red, nuclei labelled with SPIF DNA stain. (A) Wild-type egg chamber at stage 10, just prior to formation of cytoplasmic actin fibers. (B) sqh2; hs-sqh+ egg chamber, 3 days without RLC induction, which has progressed to stage 10B-11 based on the presence of cytoplasmic actin fibers (act). Follicle cells have proliferated normally. Centripetal cells (cc) elongate at stage 10 in wild-type chambers (A), but fail to elongate in sqh2 chambers (B). Wild-type border cells (bc) reach the oocyte by stage 10 (A); sqh2 border cells remain at the anterior tip of the egg chamber (B). Nuclei appear to be clogging some ring canals (B), but unlike the cell migration phenotypes, this nuclear phenotype is not seen consistently. Bar, 50 μm.

In contrast, the centripetal cells that arise from these oocyte follicle cells are very sensitive to RLC depletion. After 3 days without RLC induction, sqh2 centripetal cells fail to elongate (Fig. 5). All mature eggs display the ‘open chorion’ phenotype that results when the centripetal cells are not in position to secrete chorion material over the anterior end of the oocyte (Fig. 6). In addition, no mature eggs reach full size and most are still associated with nurse cells, indicating that dumping has failed (Fig. 6; Mahajan-Miklos and Cooley, 1994). Wheatley et al. (1995) independently identified this ‘dumpless’ phenotype in sqh1 germline clones. Thus, RLC is required in both the germ line and soma, but for mechanistically distinct processes: nurse cell contraction and follicle cell migration.

Fig. 6.

Anterior chorion formation is disrupted in sqh eggs, but rescued by RLC induction after stage 10. DIC micrographs show eggs from sqh2; hs-sqh+ mothers raised to eclosion on HS. Arrows point to dorsal appendages, anterior is left. (A-C) Typical mature eggs were solubilized in 0.2% Triton X-100 to detect gaps in the chorion. Eggs were collected: (A) 3 days, (B) 4 days, or (C) 1 day after the mother was last heat shocked. (C) Rescued egg, with normal, elongated dorsal appendages and a sealed chorion (ooplasm is not extracted by Triton solution). (A,B) RLC-depleted eggs have short, fan shaped dorsal appendages and the anterior chorion has completely failed to close, allowing the ooplasm to be extracted. (D,E) Mature eggs from sqh mothers, 3 days without heat shock followed by one heat shock and a one day recovery. Eggs show failure of dumping from early deprivation of RLC, but rescue of dorsal appendage elongation because RLC is supplied at the time of dorsal appendage follicle cell migration. (D) ‘Dumpless’ egg chamber indicates that dorsal appendage follicle cells can migrate even when nurse cells retain most of their cytoplasm (bracket); (E) more typical egg with partial rescue of dumping, though as in D the remaining nurse cell material (bracket) does not prevent rescue of dorsal appendages. Bar, 100 μm.

Fig. 6.

Anterior chorion formation is disrupted in sqh eggs, but rescued by RLC induction after stage 10. DIC micrographs show eggs from sqh2; hs-sqh+ mothers raised to eclosion on HS. Arrows point to dorsal appendages, anterior is left. (A-C) Typical mature eggs were solubilized in 0.2% Triton X-100 to detect gaps in the chorion. Eggs were collected: (A) 3 days, (B) 4 days, or (C) 1 day after the mother was last heat shocked. (C) Rescued egg, with normal, elongated dorsal appendages and a sealed chorion (ooplasm is not extracted by Triton solution). (A,B) RLC-depleted eggs have short, fan shaped dorsal appendages and the anterior chorion has completely failed to close, allowing the ooplasm to be extracted. (D,E) Mature eggs from sqh mothers, 3 days without heat shock followed by one heat shock and a one day recovery. Eggs show failure of dumping from early deprivation of RLC, but rescue of dorsal appendage elongation because RLC is supplied at the time of dorsal appendage follicle cell migration. (D) ‘Dumpless’ egg chamber indicates that dorsal appendage follicle cells can migrate even when nurse cells retain most of their cytoplasm (bracket); (E) more typical egg with partial rescue of dumping, though as in D the remaining nurse cell material (bracket) does not prevent rescue of dorsal appendages. Bar, 100 μm.

The follicle cells that construct the dorsal appendages also fail to migrate properly as RLC is depleted. These cells deposit chorion material as they move away from the oocyte and thus they build their own path for migration. In sqh mutants, these cells continue to secrete the dorsal appendage chorion material, but they fail to move away from the oocyte, leaving short, broad dorsal appendages at the normal positions (Fig. 6A-C). Since sqh mutants fail to complete dumping, we wondered whether the abnormal presence of the nurse cells was blocking the dorsal appendage cells. To test this hypothesis, we induced RLC in sqh egg chambers that had already failed to complete dumping and found that dorsal appendage formation was rescued (Fig. 6D,E). Thus a shortage of RLC in the migrating cells appears to be the primary cause of the shortened dorsal appendage phenotype.

Long-term RLC depletion disrupts interfollicular stalk formation

With prolonged lack of RLC induction (>1 week), pre-stage 10 stage egg chambers begin to show increasingly severe defects. In germaria of sqh1 mutant ovaries, which produce low levels of RLC without induction, follicle cells continue to divide normally and begin to form interfollicular stalks between germline cysts. However, while wild-type interfollicular stalks have a uniform diameter of one cell (Fig. 7A), sqh1 interfollicular stalks are disorganized (Fig. 7B). After longer periods without RLC induction, almost all ovarioles contain egg chambers with excess nurse cells and multiple oocytes (Fig. 7C). Finally, only long tubes of germ cells are found in each ovariole following the germarium. In these chambers and tubes, each oocyte tends to attract its own patch of cuboidal oocyte follicle cells (Fig. 8A-C).

Fig. 7.

Confocal immunofluorescence analysis of ovaries shows RLC is required for interfollicular stalk morphogenesis and correct organization of egg chambers. (A-C) Red, staining with anti-myosin; green, F-actin labelled with rhodamine-phalloidin (yellow indicates overlap of actin and myosin); blue, nuclei labelled with SPIF DNA stain. (A) Wild-type, early stage egg chambers are connected by interfollicular stalks (s) after leaving the germarium (g). (B) sqh1; hssqh+ ovariole; 11 days since last heat shock and eclosion of mother. Follicle cells migrate between egg chambers but fail to organize a proper stalk. Arrows indicate clusters of presumptive stalk cells. (C) sqh1; hs-sqh+ egg chamber, 16 days since last heat shock and eclosion of mother. Chamber has several times the normal number of nurse cells and 4 oocytes are visible (*, identified by their intense phalloidin staining, four ring canals and lack of a polyploid nucleus). (D) sqh2; hs-sqh+ ovariole; 16 days since last heat shock and eclosion of mother (rare survivor), labelled as in Fig. 5. Since the stronger sqh2 allele prevents follicle cell cytokinesis, no follicle cell layer appears below the muscle sheath (arrowhead). Bars, 50 μm (B and C, same scale).

Fig. 7.

Confocal immunofluorescence analysis of ovaries shows RLC is required for interfollicular stalk morphogenesis and correct organization of egg chambers. (A-C) Red, staining with anti-myosin; green, F-actin labelled with rhodamine-phalloidin (yellow indicates overlap of actin and myosin); blue, nuclei labelled with SPIF DNA stain. (A) Wild-type, early stage egg chambers are connected by interfollicular stalks (s) after leaving the germarium (g). (B) sqh1; hssqh+ ovariole; 11 days since last heat shock and eclosion of mother. Follicle cells migrate between egg chambers but fail to organize a proper stalk. Arrows indicate clusters of presumptive stalk cells. (C) sqh1; hs-sqh+ egg chamber, 16 days since last heat shock and eclosion of mother. Chamber has several times the normal number of nurse cells and 4 oocytes are visible (*, identified by their intense phalloidin staining, four ring canals and lack of a polyploid nucleus). (D) sqh2; hs-sqh+ ovariole; 16 days since last heat shock and eclosion of mother (rare survivor), labelled as in Fig. 5. Since the stronger sqh2 allele prevents follicle cell cytokinesis, no follicle cell layer appears below the muscle sheath (arrowhead). Bars, 50 μm (B and C, same scale).

Fig. 8.

Distribution of cytoskeletal proteins in sqh1; hssqh+ ovaries. (A-C) Upper surface of a bipolar egg chamber from sqh mother, 11 days since last heat shock and eclosion, triple stained for myosin (A), actin (B) and DNA (C) as in Fig. 7. Despite the presence of binucleate cells (arrows in B, C), the follicular epithelium has migrated over each of the chamber’s two oocytes (upper and lower ends, determined by viewing other optical sections, not shown). Most myosin is an abnormal, aggregated state (A), but actin is rarely detectable in the myosin aggregates (B). Bar, 25 μm. (D) Germarium region from sqh mother, 13 days since last heat shock and eclosion, stained with α spectrin antiserum. Early germ cell divisions remain normal, as evidenced by the presence of fusomes/spectrosomes in each cyst. In early cysts spectrin concentrates in the fusome precursor, the spectrosome (s); later, fusomes acquire a spider-like morphology (f) then spectrin becomes apparent at the membrane (m), its typical location (Lin and Spradling, 1995). Bar, 10 μm.

Fig. 8.

Distribution of cytoskeletal proteins in sqh1; hssqh+ ovaries. (A-C) Upper surface of a bipolar egg chamber from sqh mother, 11 days since last heat shock and eclosion, triple stained for myosin (A), actin (B) and DNA (C) as in Fig. 7. Despite the presence of binucleate cells (arrows in B, C), the follicular epithelium has migrated over each of the chamber’s two oocytes (upper and lower ends, determined by viewing other optical sections, not shown). Most myosin is an abnormal, aggregated state (A), but actin is rarely detectable in the myosin aggregates (B). Bar, 25 μm. (D) Germarium region from sqh mother, 13 days since last heat shock and eclosion, stained with α spectrin antiserum. Early germ cell divisions remain normal, as evidenced by the presence of fusomes/spectrosomes in each cyst. In early cysts spectrin concentrates in the fusome precursor, the spectrosome (s); later, fusomes acquire a spider-like morphology (f) then spectrin becomes apparent at the membrane (m), its typical location (Lin and Spradling, 1995). Bar, 10 μm.

We attempted to learn the origin of the large egg chambers by examining early cysts. The initial formation and separation of cysts remains nearly normal from one to two weeks without RLC induction as assayed by actin and spectrin localization. In both wild-type and sqh1 germaria, spectrin shows the same dynamic pattern of localization (Lin et al., 1994). However, compared to wild type, sqh1 germaria contain roughly twice as many early cysts and ‘lens-shaped’ cysts are rare (Fig. 8D). These results suggest that, in sqh1 ovaries, cyst formation is normal, but partitioning of cysts by follicle cells is slowed. Beyond 8 days after the last RLC induction, most sqh1 ovarioles have both early stage, normally partitioned chambers and later stage, fused chambers. Thus, the fusion must occur between cysts that were initially separated. This suggests that the defective interfollicular stalks fail to maintain the separation between the chambers and allow them to fuse back together as they grow.

In the most severely affected sqh egg chambers that we can generate, the follicle cell layer is entirely missing. In sqh2 mutants, which produce little or no RLC, follicle cell cytokinesis eventually fails, resulting in a few large polyploid follicle cells that cannot cover the egg chamber. These chambers are extremely disrupted, although the ring canals (not shown, but similar to Fig. 7C) and overlying muscle sheath (Fig. 7D) are apparently normal.

Myosin heavy chain forms inactive aggregates in sqh tissues

We used immunofluorescence to examine the distribution of myosin in RLC-depleted tissue. sqh2 cells are missing the uniform cortical myosin staining seen in wild-type cells (Fig. 3A,B,D,F versus Fig. 9A), while, in sqh1 cells, the cortical staining is detectable but greatly reduced (Fig. 8A). Instead, we find brightly staining clumps and speckles of various size distributed around the cytoplasm and perinuclear region (Figs 7B,C, 8A, 9A). Western blotting shows that equal concentrations of intact MHC are present in sqh1 and wild-type tissues (X.-j. Chang, K. A. E. and D. P. K., unpublished). These clumps must therefore represent a precipitated form of myosin (see Discussion).

Fig. 9.

p127 does not colocalize with RLC-deficient myosin. Egg chambers from a single batch of sqh2; hs-sqh+ females (4 days since last heat shock; 3 days since eclosion) were fixed and split into two samples. Confocal sections of typical stage 10 follicle cells in the region of the oocyte/nurse cell boundary are shown for each sample. (A) Sample stained using anti-myosin shows myosin aggregates in the cytoplasm, in place of the cortical staining that normally outlines cells (Fig. 3). (B) Sample stained using anti-p127. The p127 protein remains at the lateral plasma membranes in sqh as it does in wildtype tissues (Strand et al., 1994a). It is not detectable in myosin aggregates. Frequent occurrence of binucleate cells (arrow), due to failure of cytokinesis, indicates severe loss of myosin function. Antibody images are shown at high gain to make visible the low levels of cytoplasmic protein and background staining. Bar, 10 μm.

Fig. 9.

p127 does not colocalize with RLC-deficient myosin. Egg chambers from a single batch of sqh2; hs-sqh+ females (4 days since last heat shock; 3 days since eclosion) were fixed and split into two samples. Confocal sections of typical stage 10 follicle cells in the region of the oocyte/nurse cell boundary are shown for each sample. (A) Sample stained using anti-myosin shows myosin aggregates in the cytoplasm, in place of the cortical staining that normally outlines cells (Fig. 3). (B) Sample stained using anti-p127. The p127 protein remains at the lateral plasma membranes in sqh as it does in wildtype tissues (Strand et al., 1994a). It is not detectable in myosin aggregates. Frequent occurrence of binucleate cells (arrow), due to failure of cytokinesis, indicates severe loss of myosin function. Antibody images are shown at high gain to make visible the low levels of cytoplasmic protein and background staining. Bar, 10 μm.

Native Drosophila myosin likely binds several proteins in the cell, but only two are characterized: p127 encoded by the lethal(2)giant larvae tumor suppressor gene (Strand et al., 1994b) and F-actin. To test whether the myosin clumps might directly interfere with the cytoskeleton, we analyzed the distribution of these proteins in sqh mutants. In follicle cells, p127 staining is normally concentrated at the lateral membranes (Strand et al., 1994a). In sqh tissue, p127 shows the same pattern, with no punctate cytoplasmic staining (Fig. 9B). Factin, assayed by phalloidin staining, also retains its normal distribution and does not stain the clumps (Fig. 8A,B), except in rare cases (occurring less than once per egg chamber) when a clump becomes very large (data not shown). α-spectrin also shows normal staining in sqh ovaries (Fig. 8D). Thus sqh mutations cause mislocalization of myosin, but not of three other major cytoskeletal proteins.

In this report, we newly identify six morphogenetic processes that require myosin II-driven contraction. Nonmuscle myosin II forms bipolar filaments which, by analogy to muscle myosin, are thought to drive contractile events in the cell by drawing together actin filaments of opposite polarity. When these actin filaments are attached to the cortex or plasma membrane, the forces produced by myosin can alter the shape of the cell and neighboring cells that adhere to it.

We find that in order to carry out its diverse developmental roles, myosin is deployed in a remarkable variety of functional arrangements that would each have distinct force-generating characteristics (Table 1). These arrangements, discussed individually below, include: (1) a purse string-like band of actomyosin around the apex of each cell, described by Condic et al. (1991), that contributes to imaginal disc eversion; (2) a supracellular actomyosin band running through the edge of a cell sheet, as seen in the centripetal cells; and (3) an unexpected accumulation of actomyosin at the leading edge of locomoting border cells. We also confirm the previously reported roles for myosin in cytokinesis and nurse cell cytoplasm transport. These processes use subcellular arrangements of myosin distinct from each other and from those listed above (DeLozanne and Spudich, 1987; Karess et al., 1991; Wheatley et al., 1995).

Table 1.

Summary of developmentally specific functions for Drosophila nonmuscle myosin II

Summary of developmentally specific functions for Drosophila nonmuscle myosin II
Summary of developmentally specific functions for Drosophila nonmuscle myosin II

RLC depletion inactivates myosin

We used mutations in sqh to reduce or eliminate production of myosin’s RLC subunit. An inducible RLC transgene can temporarily rescue this defect, allowing RLC accumulation to be reduced in a reversible fashion at specific developmental stages. In sqh mutant tissues, the RLC-deficient MHC is not degraded, but instead leaves its normal location in the cortex and accumulates into cytoplasmic clumps. Several recent reports suggest a reasonable explanation for this effect. Crystallographic analyses of muscle myosins reveal that RLC wraps around an alpha helical portion of MHC at the head-tail junction, burying a large hydrophobic patch on MHC (Rayment et al., 1993; Xie et al., 1994). Exposure of this patch appears to cause precipitation of RLC-deficient myosins, both in vivo and in vitro. MHC aggregates are seen in RLC null Dictyostelium cells (Chen et al., 1994), but removal of the hydrophobic patch restores myosin function (Uyeda and Spudich, 1993). Electron microscopy shows that RLC-stripped smooth muscle MHC aggregates at the region of the hydrophobic patch (Trybus et al., 1994). The sqh MHC clumps are likely to be equivalent to these aggregates seen in other systems.

The sqh phenotypes described here could be caused either by loss of myosin activity at the cortex, or by a deleterious new activity associated with the MHC clumps. We think the latter is very unlikely for several reasons. First, the MHC clumps appear to be devoid of F-actin and so cannot produce abnormal forces. Second, p127 (Strand et al., 1994b) does not appear in the clumps. Thus, neither of the two characterized Drosophila nonmuscle myosin binding proteins are sequestered by the MHC clumps. Third, after heat-shock induction of RLC is stopped, MHC clumps begin to appear before myosin is fully depleted at the cortex, but this does not lead to any defect as long as sufficient cortical myosin remains (data not shown). Finally, all sqh defects can be reasonably explained by simple loss of myosin contractile activity. Because RLC is a constitutive subunit of myosin required for its structural integrity and correct localization, sqh hypomorphic mutations alone are not informative about the role of RLC in myosin regulation. However, sqh provides a background in which to test the function of RLCs with site-directed mutations in suspected regulatory phosphorylation sites.

‘Housekeeping’ versus developmental functions

It appears that all Drosophila cells require low levels of myosin function for viability, but higher levels are required for many developmentally programmed cell shape changes. Homozygotes of the newly generated severe allele, sqh2, complete embryonic development due to the large maternal contribution of RLC to the egg, but as larvae they remain small and soon die. This indicates RLC is not only required for cytokinesis, but also for postmitotic growth and viability, which are independent of cytokinesis at the larval stage (Gatti and Baker, 1989). Similarly, clones of zipper null cells, which make no MHC, do not survive through development (Young et al., 1993; D. P. K. and R. Montague, unpublished observations). Therefore, small amounts of myosin activity must be present to allow mutant cells to survive and exhibit informative phenotypes. The sqh1 allele is well suited for this purpose because it produces enough normal RLC to support cell viability and moderate rates of cytokinesis. In a sqh2 background, residual RLC from hs-sqh+ induction can also support cytokinesis, but only for roughly 2 days. Unfortunately, because these basal requirements for myosin must be satisfied, we cannot prove that a given process does not require any myosin. When a process occurs normally in sqh animals, we can only say it requires significantly less myosin than the processes that are disrupted. For example, during oogenesis, border cell migration requires more myosin than cytokinesis does, while oocyte follicle cell migration requires less than cytokinesis (assuming heat-shock-induced RLC is produced and degraded at similar rates in these follicle cells).

Imaginal disc morphogenesis requires RLC

Distinct phenotypes appear in adults that are deprived of RLC as larvae or pupae. A severe rough eye phenotype is fully penetrant in adults deprived of RLC induction at mid-third instar larval period. At this time, dramatic cell shape changes are beginning to occur in the eye imaginal disc, which might make this tissue especially sensitive to reduced myosin function. Rapid proliferation may also dilute the RLC remaining from the last induction. If the eye defect was caused solely by failure of cytokinesis, it should also result when younger larvae are deprived of RLC induction. Since it does not, we suspect the rough eye phenotype may reflect additional roles for myosin in one or more specific aspects of eye disc morphogenesis, which may be revealed by a more elaborate phenotypic analysis.

The shortened leg phenotype (mlf), most common in adults deprived of RLC at pupariation, can be interpreted as a failure of a specific contractile cell shape change during leg morphogenesis. In late third instar larvae, the leg is a flattened epithelial sac; at pupariation, a peak of ecdysone signals the disc to telescope outward to reach the dimensions of the mature leg (Fristrom and Fristrom, 1993). This process of disc eversion is accompanied, and apparently driven, by changes in the dimensions of individual cells (Condic et al., 1991). Before eversion, the leg disc cells are about three-four times longer around the circumference of the disc than along its proximal-distal axis. During eversion these cells become isometric, causing the disc diameter to shrink and its height to increase. This change could be accomplished simply by allowing the cortical actomyosin in each cell to contract (Condic et al., 1991). In mutants that underproduce the Broad-Complex transcription factors and therefore misregulate a number of downstream genes, this cell/tissue shape change is hindered and the mlf phenotype results. This phenotype is dominantly enhanced by several mutations, including some zip (MHC) alleles (von Kalm et al., 1995; Fristrom and Fristrom, 1993). Using partial rescue of sqh, we find reduction of myosin activity alone is sufficient to yield an identical phenotype. Thus, our results provide direct evidence that myosin drives leg disc eversion since, as predicted by the cortical contraction model, high levels of myosin function are required at the time of disc eversion for this process to occur. Myosin-powered cell shape changes that are autonomous, yet concerted, are likely to be a commonly used mechanism in epithelial morphogenesis (Young et al., 1991; Leptin et al., 1992).

Several distinct cell movements in oogenesis require RLC

The somatic follicle cells that cover each germ cell cyst form a simple cuboidal epithelium in early stage egg chambers. The follicle cells then become divided into populations that undergo a series of four migrations. Our results provide evidence that each migration occurs by a different mechanism and that three of the four migrations rely heavily on myosin. First, the border cells migrate through the center of the nurse cell cluster to the anterior face of the oocyte. Myosin is normally concentrated at both the leading and trailing edges of the border cells and is required for their crawling movement (Figs 3B-E, 5). The border cells offer an intriguing model system to study the mechanism and regulation of directed, invasive cell movements (Montell et al., 1992). Second, most of the remaining epithelium shifts over the nurse cells to contact the oocyte (Fig. 3E), even in sqh mutants with little functional myosin (Fig. 8B). This movement could be driven, for example, by changes in cell adhesion or by motors other than myosin II; or it may simply use such small amounts of myosin II that we cannot detect the requirement. Third, the centripetal cells normally close over the anterior face of the oocyte like an iris and subsequently form the anterior portion of the eggshell. In sqh2 mutants, this movement fails (see below). Fourth, the migration of dorsal appendage follicle cells is strongly inhibited in sqh mutants, even though these cells locomote over a free surface, provided by the chorion they secrete. Unfortunately, the chorion defeats our attempts to observe myosin distribution and shape changes in these cells.

An actomyosin contractile band drives centripetal cell migration

The ring of centripetal cells develops a distinct distribution of myosin that can account for its movement: myosin and actin accumulate at the leading edge of each ingressing cell, resulting in a contractile band that is essentially continuous around the egg chamber. Contraction of this actomyosin band would pull the centripetal cells inward, much like the contractile ring pulls the cleavage furrow of a dividing cell although, in this case, the actomyosin band is interrupted by cell-cell contacts. For this mechanism to work, the centripetal cells must adhere more strongly to each other than to the surrounding germ cells. According to this model, the centripetal cells could not migrate autonomously, but only as an intact epithelium, a prediction that may be testable using genetic mosaics.

In sqh egg chambers, myosin is removed from the band and the centripetal cells fail to move, proving that RLC is required for this process. This phenotype is analogous to the failure of dorsal closure that characterizes zip null embryos, which lack another myosin subunit, MHC (Young et al., 1993). Centripetal cell migration and dorsal closure are strikingly similar to epithelial wound healing, which is also associated with a purse stringlike myosin band at the cells’ leading edge (Bement et al., 1993). Our results confirm the importance of myosin in morphogenetic movements that resemble wound healing, and suggest that purse string-like contraction may be a general tactic used in metazoan development. Several Drosophila genes have been found to be expressed specifically in either the centripetal cells, or the leading edge cells during dorsal closure. For example, dpp, which encodes a TGF-β-like signalling molecule, is expressed in both cell types (R. W. Padgett, personal communication and Twombly et al., 1996). Similarly, mammalian TGFβ1 is expressed in leading edge cells during embryonic wound healing (Martin et al., 1993). Genes such as these might have a role in organizing or activating myosin specifically when a purse-string configuration is required.

Contrasting roles for myosin in Dictyostelium and Drosophila

Multicellular development evolved independently in Dictyostelium and metazoans such as Drosophila. Thus it is useful to compare the roles of myosin in the two organisms. Springer et al. (1994) employed a cold-sensitive myosin mutation to show that Dictyostelium fruiting body morphogenesis requires myosin at two distinct times: early, for correct cell sorting within the mound and, late, to generate an elongated stalk. The specific cellular basis of the late myosin requirement is unclear. However, the early requirement can be explained by the fact that myosin null cells become immobile once they aggregate into a mound (Doolittle et al., 1995). This prevents the autonomous migrations by which the cells sort to their correct locations and allow fruiting body formation to proceed. It is suggested that migration halts at the mound stage because the myosin null cells cannot overcome adhesion to their neighbors (Doolittle et al., 1995).

This phenotype is similar to the failure of border cell migration in sqh egg chambers. Both defects may reflect an ancient role for myosin II in amoeboid cell motility. However, the centripetal cells in the egg chamber and leading edge cells during dorsal closure appear to use myosin in a fundamentally different way to migrate. These cells maintain a polarized epithelial sheet as they move. Myosin is recruited to a band along one edge of the sheet and the subsequent elongation of the sheet can be explained by a purse-string-like contraction of the myosin band. Thus, Dictyostelium cells appear to use myosin autonomously to counteract adhesive forces while, in Drosophila, cells can move nonautonomously, using adhesion to effectively distribute the forces generated by myosin.

Summary of myosin-based processes in Drosophila

Using a variety of genetic and cell biological approaches, an extensive catalog of Drosophila myosin functions has been assembled (Table 1). It is apparent from this table that myosin II is extremely versatile, participating in assorted types of cell shape changes and localizing in several discreet patterns. Remarkably, none of the five or more unconventional myosins expressed in these cells (Morgan et al., 1994), or any other motors, can functionally substitute for myosin II in these cases. This list suggests myosin II may be the primary contractileforce producing protein in development. However, very little myosin is required in many other situations, including actinbased processes such as bristle elongation and nurse cell actin cable formation, cell movements such as oocyte follicle cell migration and all cell fate decisions that we have assayed. This specificity suggests that genetic approaches to myosin function and regulation will continue to be profitable.

Table 1 also provides a checklist of tests that can be used to characterize the biological activity of site-directed mutant RLCs. For example, we can test whether any of the multiple phosphorylation sites on RLC are dedicated to regulating myosin function during specific developmental processes. If so, an RLC with a mutation at that site might support some of these processes but not others. Our preliminary studies show the myosin light chain kinase phosphorylation site may have this kind of specificity (unpublished observations). Also, genes that can mutate to give a subset of the phenotypes in Table 1 (or alter the corresponding myosin structures) would be good candidates for myosin regulators or interactors. For example, expression of a dominant inhibitory Rac eliminates the myosin purse string that normally forms at dorsal closure (Harden et al., 1995). Given the intensity with which many signalling molecules are being studied in Drosophila, we may soon be able to describe a complete pathway from cell determination to morphogenesis.

We thank Madelyn Demsky and Richard Lin for technical assistance, L. Eibest for scanning electron microscopy, B. Mechler for kindly providing anti-p127 antibodies, members of the Kiehart Lab for support and helpful discussions, Drs Meg Titus, Haifan Lin, Rick Fehon, and Dianne and James Fristrom for key insights. Supported by NIH GM33830, the March of Dimes and NIH predoctoral fellowship GM07184 to K. A. E.

Ashburner
,
M.
(
1989
).
Drosophila: A Laboratory Handbook
.
Cold Spring Harbor, NY
:
Cold Spring Harbor Laboratory Press
.
Bement
,
W. M.
,
Forscher
,
P.
and
Mooseker
,
M. S.
(
1993
).
A novel cytoskeletal structure involved in purse string wound closure and cell polarity maintenance
.
J. Cell Biol
.
121
,
565
578
.
Chen
,
P.
,
Ostrow
,
B. D.
,
Tafuri
,
S. R.
and
Chisholm
,
R. L.
(
1994
).
Targeted disruption of the Dictyostelium RMLC gene produces cells defective in cytokinesis and development
.
J. Cell Biol
.
127
,
1933
1944
.
Cheney
,
R. E.
,
Riley
,
M. A.
and
Mooseker
,
M. S.
(
1993
).
Phylogenetic analysis of the myosin superfamily
.
Cell Motility and the Cytoskeleton
24
,
215
233
.
Condic
,
M. L.
,
Fristrom
,
D.
and
Fristrom
,
J. W.
(
1991
).
Apical cell shape changes during Drosophila imaginal leg disc elongation: a novel morphogenetic mechanism
.
Development
111
,
23
33
.
DeLozanne
,
A.
and
Spudich
,
J.
(
1987
).
Disruption of the Dictyostelium myosin heavy chain gene by homologous recombination
.
Science
236
,
1086
1091
.
Doolittle
,
K. W.
,
Reddy
,
I.
and
McNally
,
J. G.
(
1995
).
3D analysis of cell movement during normal and myosin-II-null cell morphogenesis in Dictyostelium
.
Dev. Biol
.
167
,
118
129
.
Edwards
,
K. A.
,
Chang
,
X.-j.
and
Kiehart
,
D. P.
(
1995
).
Essential light chain of Drosophila nonmuscle myosin II
.
J. Muscle Research & Cell Motility
16
,
491
498
.
Fristrom
,
D.
and
Fristrom
,
J. W.
(
1993
).
The metamorphic development of the adult epidermis
.
In The Development of Drosophila melanogaster
(ed.
M.
Bate
and
A.
Martinez-Arias
.), pp.
843
897
.
Cold Spring Harbor, NY
:
Cold Spring Harbor Laboratory Press
.
Gatti
,
M.
and
Baker
,
B. S.
(
1989
).
Genes controlling essential cell-cycle functions in Drosophila melanogaster
.
Genes Dev
.
3
,
438
453
.
Hall
,
A.
(
1994
).
Small GTP-Binding proteins and the regulation of the actin cytoskeleton
.
Annu. Rev. Cell Biol
.
10
,
31
54
.
Harden
,
N.
,
Loh
,
H. Y.
,
Chia
,
W.
and
Lim
,
L.
(
1995
).
A dominant inhibitory version of the small GTP-binding protein Rac disrupts cytoskeletal structures and inhibits developmental cell shape changes in Drosophila
.
Development
121
,
903
914
.
Harrington
,
W. F.
and
Rodgers
,
M. E.
(
1984
).
Myosin
.
Annu. Rev Biochem
.
53
,
35
73
.
Karess
,
R. E.
,
Chang
,
X.-j.
,
Edwards
,
K. A.
,
Kulkarni
,
S.
,
Aguilera
,
I.
and
Kiehart
,
D. P.
(
1991
).
The regulatory light chain of nonmuscle myosin is encoded by spaghetti squash, a gene required for cytokinesis in Drosophila
.
Cell
65
,
1177
1189
.
Ketchum
,
A. S.
,
Stewart
,
C. T.
,
Stewart
,
M.
and
Kiehart
,
D. P.
(
1990
).
Complete sequence of the Drosophila nonmuscle myosin heavy chain transcript: Conserved sequences in the myosin tail and differential splicing in the 5′ untranslated sequence
.
Proc. Natl. Acad. Sci. USA
87
,
6316
6320
.
Kiehart
,
D. P.
and
Feghali
,
R.
(
1986
).
Cytoplasmic myosin from Drosophila melanogaster
.
J. Cell Biol
.
103
,
1517
1525
.
Kiehart
,
D. P.
,
Lutz
,
M. S.
,
Chan
,
D.
,
Ketchum
,
A. S.
,
Laymon
,
R. A.
,
Nguyen
,
B.
and
Goldstein
,
L. S. B.
(
1989
).
Identification of the gene for fly non-muscle myosin heavy chain: Drosophila myosin heavy chains are encoded by a gene family
.
EMBO J
8
,
913
922
.
Kiehart
,
D. P.
,
Ketchum
,
A.
,
Young
,
P.
,
Lutz
,
D.
,
Alfenito
,
M. R.
,
Chang
,
X.-j.
,
Awobuluyi
,
M.
,
Pesacreta
,
T. C.
,
Inoue
,
S.
,
Stewart
,
C. T.
and
Chen
,
T.-l.
(
1990
).
Contractile proteins in Drosophila development
.
Ann. N.Y. Academy Sci
.
582
,
233
251
.
Leptin
,
M.
,
Casal
,
J.
,
Grunewald
,
B.
and
Reuter
,
R.
(
1992
).
Mechanisms of early Drosophila mesoderm formation
.
In Gastrulation
(ed.
C.
Stern
and
P.
Ingham
) Development
1992
Supplement
,
23
31
.
Lin
,
H.
,
Yue
,
L.
and
Spradling
,
A. C.
(
1994
).
The Drosophila fusome, a germline-specific organelle, contains membrane skeletal proteins and functions in cyst formation
.
Development
120
,
947
956
.
Lin
,
H.
and
Spradling
,
A. C.
(
1995
).
Fusome asymmetry and oocyte determination in Drosophila
.
Dev. Genet
.
16
,
6
12
.
Lindsley
,
D. L.
and
Zimm
,
G. G.
(
1992
).
The Genome of Drosophila melanogaster
.
New York, NY
:
Academic Press, Inc. (Harcourt Brace Jovanovich
).
Lundell
,
M. J.
and
Hirsh
,
J.
(
1994
).
A new visible light DNA fluorochrome for confocal microscopy
.
Biotechniques
16
,
434
440
.
Mabuchi
,
I.
and
Okuno
,
M.
(
1977
).
The effect of myosin antibody on the division of starfish blastomeres
.
J Cell Biol
.
74
,
251
263
.
Mahajan-Miklos
,
S.
and
Cooley
,
L.
(
1994
).
Intercellular cytoplasm transport during Drosophila oogenesis
.
Dev. Biol
.
165
,
336
351
.
Mansfield
,
S. G.
,
Al-Shirawi
,
D. Y.
,
Ketchum
,
A. S.
,
Newbern
,
E. C.
and
Kiehart
,
D. P.
(
1996
).
Molecular organization and alternative splicing in zipper, the gene that encodes the Drosophila nonmuscle myosin II heavy chain
.
J. Mol. Biol
.
255
,
98
109
.
Martin
,
P.
,
Dickson
,
M. C.
,
Millan
,
F. A.
and
Akhurst
,
R. J.
(
1993
).
Rapid induction and clearance of TGF β1 is an early response to wounding in the mouse embryo
.
Dev Genet
.
14
,
225
238
.
Montell
,
D. J.
,
Rorth
,
P.
and
Spradling
,
A. C.
(
1992
).
slow border cells, a locus required for a developmentally regulated cell migration during oogenesis, encodes Drosophila C/EBP
.
Cell
71
,
51
62
.
Morgan
,
N. S.
,
Skovronsky
,
D. M.
,
Artavanis-Tsakonas
,
S.
and
Mooseker
,
M. S.
(
1994
).
The molecular cloning and characterization of Drosophila melanogaster Myosin-IA and Myosin-IB
.
J. Mol. Biol
.
239
,
347
356
.
Pesacreta
,
T. C.
,
Byers
,
T. J.
,
Dubreuil
,
R.
,
Kiehart
,
D. P.
and
Branton
,
D.
(
1989
).
Drosophila spectrin: the membrane skeleton during embryogenesis
.
J. Cell Biol
.
108
,
1697
1709
.
Rayment
,
I.
,
Rypniewski
,
W. R.
,
Schmidt-Base
,
K.
,
Smith
,
R.
,
Tomchick
,
D. R.
,
Benning
,
M. M.
,
Winkelmann
,
D. A.
,
Wesenberg
,
G.
and
Holden
,
H. M.
(
1993
).
Three-dimensional structure of myosin subfragment-1: A molecular motor
.
Science
261
,
50
58
.
Robertson
,
H. M.
,
Preston
,
C. R.
,
Phillis
,
R. W.
,
Johnson-Schlitz
,
D. M.
,
Benz
,
W. K.
and
Engels
,
W. R.
(
1988
).
A stable genomic source of P-element transposase in Drosophila melanogaster
.
Genetics
118
,
461
470
.
Spradling
,
A. C.
(
1993
).
Developmental genetics of oogenesis
.
In The Development of Drosophila melanogaster
(ed.
M.
Bate
and
A.
Martinez-Arias
.), pp.
1
70
.
Cold Spring Harbor, NY
:
Cold Spring Harbor Laboratory Press
.
Springer
,
M. L.
,
Patterson
,
B.
and
Spudich
,
J. A.
(
1994
).
Stage-specific requirement for myosin II during Dictyostelium development
.
Development
120
,
2651
2660
.
Strand
,
D.
,
Raska
,
I.
and
Mechler
,
B. M.
(
1994a
).
The Drosophila lethal(2)giant larvae tumor suppressor protein is a component of the cytoskeleton
.
J. Cell Biol
.
127
,
1345
1360
.
Strand
,
D.
,
Jakobs
,
R.
,
Merdes
,
G.
,
Neumann
,
B.
,
Kalmes
,
A.
,
Heid
,
H. W.
,
Husmann
,
I.
and
Mechler
,
B. M.
(
1994b
).
The Drosophila lethal(2)giant larvae tumor suppressor forms homo-oligomers and is associated with nonmuscle myosin II heavy chain
.
J. Cell Biol
.
127
,
1361
1373
.
Thummel
,
C. S.
and
Pirotta
,
V.
(
1992
).
New pCaSpeR P element vectors
.
Dros. Inf. Serv
.
71
,
150
.
Trybus
,
K. M.
,
Waller
,
G. S.
and
Chatman
,
T. A.
(
1994
).
Coupling of ATPase activity and motility in smooth muscle myosin is mediated by the regulatory light chain
.
J. Cell Biol
.
124
,
963
969
.
Twombly
,
V.
,
Blackman
,
R. K.
,
Jin
,
H.
,
Graff
,
J. M.
,
Padgett
,
R. W.
and
Gelbart
,
W. M.
(
1996
).
The TGF-β signaling pathway is essential for Drosophila oogenesis
.
Development
122
,
1555
1565
.
Uyeda
,
T. Q. P.
and
Spudich
,
J. A.
(
1993
).
A functional recombinant myosin II lacking a regulatory light chain-binding site
.
Science
262
,
1867
1870
.
von Kalm
,
L.
,
Fristrom
,
D.
and
Fristrom
,
J.
(
1995
).
The making of a fly leg: a model for epithelial morphogenesis
.
BioEssays
17
,
693
702
.
Warrick
,
H. M.
and
Spudich
,
J. A.
(
1987
).
Myosin structure and function in cell motility
.
Annu. Rev. Cell Biol
.
3
,
379
421
.
Wheatley
,
S.
,
Kulkarni
,
S.
and
Karess
,
R.
(
1995
).
Drosophila nonmuscle myosin II is required for rapid cytoplasmic transport during oogenesis and for axial nuclear migration in early embryos
.
Development
121
,
1937
1946
.
Wolff
,
T
and
Ready
,
D. F.
(
1993
).
Pattern formation in the Drosophila retina
.
In The Development of Drosophila melanogaster
(ed.
M.
Bate
and
A.
Martinez-Arias
.), pp.
1277
1325
.
Cold Spring Harbor, NY
:
Cold Spring Harbor Laboratory Press
.
Xie
,
X.
,
Harrison
,
D. H.
,
Schlichting
,
I.
,
Sweet
,
R. M.
,
Kalabokis
,
V. N.
,
Szent-Gyorgyi
,
A. G.
and
Cohen
,
C.
(
1994
).
Structure of the regulatory domain of scallop myosin at 2.8 angstrom resolution
.
Nature
368
,
306
312
.
Young
,
P. E.
,
Pesacreta
,
T. C.
and
Kiehart
,
D. P.
(
1991
).
Dynamic changes in the distribution of cytoplasmic myosin during Drosophila embryogenesis
.
Development
111
,
1
14
.
Young
,
P. E.
,
Richman
,
A. M.
,
Ketchum
,
A. S.
and
Kiehart
,
D. P.
(
1993
).
Morphogenesis in Drosophila requires nonmuscle myosin heavy chain function
.
Genes Dev
.
7
,
29
41
.