Profilin mutations disrupt multiple actin-dependent processes during Drosophila development

The analysis of cytoskeletal components has been facilitated by the use of mutational and genetic analyses. This approach has been a successful way to reconcile and support in vitro data with in vivo observations. One difficulty often encountered in such studies is that mutagenizing cytoskeletal proteins generally leads to death. We have used oogenesis in Drosophila as a model system in which to dissect genetically the components of the actin-dependent process of cytoplasm transport. This strategy has identified mutations that affect the dynamics of the actin cytoskeleton, without affecting the viability of the fly (Cooley et al., 1992; Xue and Cooley, 1993; Cant et al., 1994). We have previously reported that the chickadee (chic) gene encodes Drosophila profilin and have described female sterile chic mutations that block polymerization of extensive networks of cytoplasmic actin filaments in the nurse cells of the egg chamber (Cooley et al., 1992).

Extensive biochemical analysis of profilin has led to several models of the physiological functions of the protein. Profilin is a 12–15×10³M_r ubiquitous cytoplasmic protein that binds actin monomers in a 1:1 complex and regulates filament polymerization (Stossel et al., 1985; Pollard and Cooper, 1986). Profilin binding to actin monomers prevents spontaneous nucleation of actin filaments (Pollard and Cooper, 1984). This prevention of nucleation by profilin does not affect elongation of existing filaments (Tilney et al., 1983; Pollard and Cooper, 1984). Profilin may also affect actin polymerization through the stimulation of nucleotide and divalent cation exchange on actin monomers (Mockrin and Korn, 1980; Nishida, 1985; Goldschmidt-Clermont et al., 1991). Since ATP-actin monomers add to growing filaments at a faster rate than ADP-actin monomers and have a lower critical concentration (Pollard, 1986), it has been proposed that in the presence of excess ATP, the stimulation of nucleotide exchange by profilin may accelerate actin polymerization (Goldschmidt-Clermont et al., 1991).

In addition to regulating actin assembly, both in vitro and in vivo evidence suggest a role for profilin in signal transduction. Profilin binds a membrane phospholipid, phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P₂), with high affinity and can regulate its hydrolysis by unphosphorylated phospholipase C-γ1 (Lassing and Lindberg, 1985; Goldschmidt-Clermont et al., 1990). The phosphorylated form of phospholipase C-γ1 can hydrolyze PtdIns(4,5)P₂ bound to profilin, presumably releasing profilin into the cytoplasm (Goldschmidt-Clermont and Janmey, 1991). In addition, evidence from Saccaromyces cerevisiae suggests a link between profilin’s regulation of PtdIns(4,5)P₂ hydrolysis and the RAS signaling pathway (Vojtek et al., 1991). Activation of the adenylyl cyclase complex by RAS requires an associated protein called CAP. Disruption of the carboxy terminus of CAP results in morphological and nutritional defects. Such mutations in CAP can be suppressed by overexpression of profilin (Vojtek et al., 1991). Analysis of profilin null mutations in yeast have provided the first genetic evidence for an important in vivo role for profilin (Magdolen et al., 1988; Haarer et al., 1990). Although profilin is non-essential in yeast under permissive growing conditions, mutant cells display a number of aberrant phenotypes. Cells lacking profilin grow more slowly, become larger than wild-type cells and are often multinucleate. In profilin mutant cells, actin cables are no longer detected and actin spots appear delocalized. Anti-actin antibodies were able to detect thick actin bars, which may have formed as a result of improper regulation of actin assembly (Haarer et al., 1990).

The chickadee gene in Drosophila contains two promoters that produce two transcripts sharing the same open reading frame encoding profilin (Cooley et al., 1992). chickadee female sterile mutations result from the disruption of the ovary-specific profilin transcript. The second transcript, that is constitutively expressed, is unaffected in homozygous mutant females. Here we report the analysis of new alleles that reduce or eliminate the constitutive expression of profilin. These new chic alleles can be organized into an allelic series of increasing severity ranging from effects on germ cell proliferation and cytokinesis, bristle and eye development to embryonic lethality in null alleles. In addition to phenotypic characterization of chickadee mutations, we raised monoclonal antibodies to Drosophila profilin to study the protein distribution and sub-cellular localization in various tissues.

All fly stocks were maintained under standard culturing conditions. Wild-type genomic DNA for Southern blot analysis was obtained from cn;ry⁵⁰⁶. Female sterility was assessed by placing homozygous females in vials with an excess of males and scoring for any progeny after 10 days. Homozygous males were tested with five virgin females and the presence of progeny was scored 10 days later. Males and females were classified as weakly fertile if they produced less than 25% of the wild-type number of progeny.

Hybridoma cell lines were derived from a mouse immunized with a fusion protein containing the profilin open reading frame cloned into the expression vector pGEX-3X (Smith and Johnson, 1988). After induction of the fusion protein in E. coli, the protein was purified from crude lysates by affinity chromatography on immobilized glutathione (Smith and Johnson, 1988). 50 μg purified fusion protein was used to immunize mice by intraperitoneal injection. Antisera were screened by western blotting to the fusion protein and immunostaining to ovaries. Monoclonal cell lines were generated following standard procedures (Harlow and Lane, 1988).

Drosophila ovaries or whole animals were ground in SDS sample buffer and protein concentration was determined using the Bio-Rad protein assay (Bio-Rad Laboratories). 15% polyacrylamide gels were loaded with 40 μg of protein per lane and transferred to Immobilon-P membrane (Millipore). Blocking, antibody incubation and detection were carried out according to Xue and Cooley (1993). Monoclonal anti-profilin cell culture supernatant from line 6F was used at a 1:10 dilution.

The procedure for antibody staining of ovaries was described previously (Xue and Cooley, 1993). Undiluted anti-profilin monoclonal supernatant from cell line 4H was used. The secondary FITC-conjugated goat anti-mouse IgG (Pierce) was used at a concentration of 1:200.

Embryos were fixed and stained according to the method of Patel et al. (1989) with the following exception. After incubation with an FITC-conjugated goat anti-mouse secondary antibody diluted 1:200, embryos were washed and mounted in 50% glycerol. Anti-profilin monoclonal supernatant from cell line 6F was used undiluted.

Ovaries were dissected in Drosophila saline solution and stained for β-galactosidase activity according to Cooley et al. (1992). For actin staining, ovaries were dissected into individual egg chambers, fixed in devitellinizing buffer and heptane (see Cooley et al., 1992), and rinsed thoroughly in saline. Rhodamine-conjugated phalloidin (10 μl) (Molecular Probes no. R-415) was vacuum dried and resuspended in 100 μl of saline. Egg chambers were incubated for 20 minutes in the dark with 50-100 μl rhodamine-phalloidin and then rinsed with saline. For DAPI staining, fixed egg chambers were incubated for 5 minutes in 1 μg/ml DAPI in saline followed by several washes in saline. Egg chambers were mounted in 50% glycerol in PBS.

To count the number of nurse cells in mutant egg chambers, either rhodamine-conjugated phalloidin was used to stain cell membranes or antibodies to the kelch protein (Xue and Cooley, 1993) were used to stain ring canals. We found that in every mutant egg chamber examined the number of ring canals corresponded to the number of nurse cells.

White prepupae were collected and the nota and heads were dissected at 38–42 hours of pupal development according to Schweisguth and Posakony (1992). After fixation, the tissue was stained with rhodamine-conjugated phalloidin as described above. The tissue was washed extensively in PBS and mounted in 50% glycerol. Bristle actin was examined on a confocal microscope where optical sections were superimposed.

Testes were dissected out of adult males in PEM (0.1 M Pipes, 1 mM MgCl₂, 1 mM EGTA pH 6.9). They were fixed in 4% paraformaldehyde in PEX (PEM + 0.1% Triton X-100) (Gönczy et al., 1992). After fixation, the testes were mounted in 50% glycerol.

Microscopy was carried out on a Zeiss Axiophot microscope equipped with differential interference contrast and epiflourescence optics. Confocal images were collected using the MRC 600 system (Bio Rad). Optical sections were combined using the program ‘Project’. For the scanning electron microscope, flies were ethanol dehydrated, mounted on stubs and air dried. Flies were viewed on an ISI-SS40 electron microscope.

To remobilize the P element in chic⁷⁸⁸⁶, a Δ2-3 jumpstarter strain (Laski et al., 1984; Robertson et al., 1988) was crossed into the chic⁷⁸⁸⁶ background to supply a source of transposase. Flies in which excision events had occurred were detected by scoring progeny for loss of the rosy⁺ eye color marker carried by the P element. The phenotype of the rosy⁻ revertants was determined by making the mutant chromosomes homozygous. Lines with phenotypes more severe than the original chic⁷⁸⁸⁶ phenotype were further characterized by genetic complementation testing and Southern blotting.

Genomic DNA for Southern blot analysis was isolated from adults using standard procedures (protocol 48 in Ashburner, 1989). DNA gels were blotted by capillary transfer to Hybond-N (Amersham) and ultraviolet cross-linked (Stratalinker, Stratagene). ³²P-labeled DNA probes were prepared using random hexamer priming according to Feinberg and Vogelstein (1983). Hybridization was in aqueous solution at 65°C with agitation (Church and Gilbert, 1984).

The expression of profilin during various stages of Drosophila development was examined by western immunoblotting (Fig. 1A). Profilin was present throughout development with a strong surge in expression in 12-to 24-hour embryos and a second less dramatic increase during pupal stages. The adult extracts showed the lowest expression levels. The reduced expression may reflect the reduced mitotic activity of adult tissue.

Because the original chic alleles were female sterile, we examined profilin in ovary extracts from wild type, chic⁷⁸⁸⁶ (a weakly fertile allele) and chic¹³²⁰ (a fully sterile allele; Fig. 1B). The amount of profilin was reduced in the mutant ovaries, with the level of reduction corresponding to the severity of the phenotype. Since female sterile alleles do not express the ovary-specific transcript (Cooley et al., 1992), the remaining protein detected in the chic¹³²⁰ lane corresponds to expression from the constitutive promoter.

The subcellular localization of profilin in egg chambers was examined, in part to determine the distribution of the residual protein in chickadee mutant ovaries. The oocyte develops in a structure called the egg chamber that is formed at the anterior of the ovary in the germarium (for oogenesis reviews, see King, 1970 and Spradling, 1994). Incomplete cytokinesis during the four mitotic divisions of a germline stem cell daughter results in sixteen germline cells interconnected by fifteen cytoplasmic bridges, or ring canals. One becomes the oocyte and the other 15 differentiate into nurse cells. After the mitotic divisions, the cluster of cells is enclosed by somatic follicle cells. The egg chambers develop in ovariole tubes, with more mature egg chambers moving posteriorly in the ovary. As the egg chamber matures, the 15 nurse cells synthesize and transport into the oocyte components that are required for subsequent embryonic development. Late in oogenesis, during stage 10B, extensive networks of actin filaments polymerize in the cytoplasm of each nurse cell. In stage 11, the nurse cells contract and their remaining cytoplasmic contents are rapidly transported into the oocyte. This results in regression of the nurse cell cluster, and doubling of the oocyte volume.

In wild-type ovaries, profilin was highly expressed in the cytoplasm of the somatic follicle cells as these cells migrate and enclose germline clusters to form egg chambers in the germarium (Fig. 2A). In these somatically derived cells, low levels of profilin were found associated with the plasma membrane (arrow in 2A). The high cytoplasmic expression in the follicle cells decreased after stage 2 (arrowhead in Fig. 2B), but the protein remained expressed in these cells throughout oogenesis (Fig. 2B). In the follicle cells covering the oocyte at stage 10 we also detected low levels of membrane associated profilin in addition to cytoplasmic profilin (not shown). Profilin was very abundantly expressed in a specialized set of follicle cells, called border cells, that migrate through the nurse cell cluster from the anterior of the egg chamber to the anterior edge of the developing oocyte (arrow in Fig. 2B). This cluster of cells was often observed extending cell surface projections during the migration. In the germarium, profilin was also expressed in the cytoplasm of the germline cells, however at a much lower level than in the somatic tissue. Profilin was detected in the nurse cell cytoplasm throughout oogenesis (Fig. 2B), with the highest level of expression at stage 10 (Fig. 2C).

Examination of ovaries from chic¹³²⁰, a strong female sterile allele, revealed that only a subset of the profilin expression pattern was affected. The expression in somatic follicle cells throughout oogenesis appeared unaffected, while expression in the nurse cell cytoplasm was dramatically reduced (Fig. 2D). This reduction was most apparent at stage 10, where the low level of profilin in the nurse cell cytoplasm correlated with the failure to polymerize extensive actin filament networks (Cooley et al., 1992). Thus, the reduction in total protein in the ovary detected by western blot (Fig. 1B) is likely to result from an absence of high levels of the protein in the nurse cell cytoplasm. These results indicate that the ovary-specific transcript that is disrupted in female sterile alleles of chickadee (Cooley et al., 1992) is in fact nurse cell-specific. It does appear that the constitutive transcript provides a basal level of profilin expression in the germline cells, which is still present in chic¹³²⁰.

To examine further the function of profilin in Drosophila we determined the phenotype of a profilin null mutation by inducing deletions at the locus using P element excision. A source of transposase was genetically introduced into an insertion allele of chickadee (chic⁷⁸⁸⁶) to catalyze the mobilization of the P element. The orientation of the P element places the rosy⁺ gene adjacent to the rest of the chickadee gene. This proved to be important for obtaining deletions extending into chickadee since we scored for excisions by selecting for loss of the rosy⁺ eye color marker. 333 independent rosy⁻ excision lines were isolated after mobilizing the P element in chic⁷⁸⁸⁶ and their phenotypes were analyzed by making the mutant chromosomes homozygous. 261 of the lines displayed homozygous phenotypes roughly the same as chic⁷⁸⁸⁶, and corresponded to internal P element excisions. The second most common event was precise excision of the P element, restoring fertility in 57 homozygous rosy⁻ lines. Eleven lines displayed phenotypes more severe than chic⁷⁸⁸⁶. Only one of these lines was homozygous viable (chic³⁷), with a severe female sterile phenotype, and the addition of male sterility and bristle defects. The other ten rosy⁻ lines were homozygous lethal.

Previously we described a divergently transcribed gene in close proximity to the chickadee locus (Cooley et al., 1992). This gene is now known to encode the Drosophila homolog of the eukaryotic translation initiation factor 4A (eIF4A; L. C. and E. M. V., data not shown; Dorn et al., 1993). While characterizing this genetic interval, we determined that mutations in the eIF4A gene are lethal and allelic to l(2)gpdh-4 (Knipple and MacIntyre, 1984). Recently, Dorn et al. (1993) described lethal mutations of eIF4A and designated the gene l(2L) 162^neo after a P element insertion mutation at the locus. We will refer to the gene encoding the eIF4A protein as eIF4A in this paper.

To establish whether the lethality of the excision lines was caused by deletion of chickadee, eIF4A or both, genetic and molecular approaches were undertaken. Complementation testing showed that the ten lethal lines fell into three groups (Fig. 3A). The group I lethals (106, 112 and 129 in Fig. 3A) failed to complement both chickadee and eIF4A mutations and are therefore deletions affecting proper functioning of both genes. These three lines were lethal in combinations with all other deletion lines obtained. The group II lethals (38, 103, 139) failed to complement the group I lethals, but partially complemented the group III lethals (chic²²¹, chic²⁸¹, 211, and 321). Complementation was scored as incomplete because trans-allelic flies were female sterile, male sterile and had bristle defects. These phenotypes were also produced by the one viable deletion line (chic³⁷) obtained from the excision screen. These data suggested that the deletions carried in the partially complementing group II and III lethals overlapped only in the region that was also deleted in chic³⁷ (Fig. 3A). The group III lethal lines failed to complement chickadee, but fully complemented the eIF4A mutant, l(2)gdh-4.

After genetically determining that the group III lethals map to chickadee, we mapped the deletion breakpoints molecularly for three of the lines using Southern blotting (Fig. 3C). Genomic DNA was prepared from chic³⁷, chic²²¹/CyO and chic²⁸¹/CyO. Breakpoints were mapped by probing for restriction fragment length polymorphisms induced by the deletion. The breakpoints for the two lethal lines, chic²²¹ and chic²⁸¹, both mapped within the fragment detected by probe I (Fig. 3C, left panel). This probe detects only chickadee transcripts on a northern blot (L. C., data not shown). The breakpoint for the viable line chic³⁷ was found using probe II, which is a small EcoRI-XbaI fragment located between the first and second exons of chickadee (Fig. 3C, center panel). The genetic complementation data with l(2)gdh-4 suggested that the eIF4A gene was probably not affected by these deletions. Probe III, which detects the DNA directly flanking the chic⁷⁸⁸⁶ P element insertion site, was used to confirm that the eIF4A gene was intact (Fig. 3C, right panel). Thus the two lethal lines chic²²¹ and chic²⁸¹ are deletions that affect only the chickadee gene, and are obligate null alleles.

Genetic analysis revealed that chickadee alleles can be arranged in a phenotypic series which correlates increasingly severe phenotypes with reduced levels of profilin expression (Table 1). The original female sterile phenotype was previously described in detail (Cooley et al., 1992) and represents loss of function for one aspect of profilin function. In subsequent sections we describe the pleiotropic consequences of reducing somatic and germline profilin concentrations using various combinations of our new chickadee alleles.

In the female sterile allele chic¹³²⁰, only the nurse cell-specific transcript was absent (Cooley et al., 1992). When this allele was made hemizygous with the null mutation chic²²¹, or a larger deficiency for the region, Df(2)GdhA, more severe egg chamber defects were observed. These were presumably due to halving the expression of the constitutive transcript in both the somatic and germline tissue. The defects were first observed by staining nuclear material with the dye DAPI. We observed variations from the normal number of fifteen nurse cell nuclei (Fig. 4D,E; Fig. 5). Often the nuclei appeared to be in pairs, so we investigated whether all the nurse cells were binucleate. Egg chambers were double stained with DAPI and rhodamine-conjugated phalloidin. Staining of the actin allowed us to map the plasma membranes of nurse cells and deduce the number of nurse cells in an egg chamber (Fig. 4A,B,C). We found that the majority of egg chambers had too few nurse cells and these were mostly binucleate, as determined by counting the number of nuclei within the egg chamber (Fig. 4D,E). The variation in cell number was great, with egg chambers containing between 2 and over 30 nurse cells (Fig. 5). Egg chambers with more than eight nurse cells had too many nuclei to count accurately. However, we were able to determine that nuclei were always in excess of nurse cell number. An additional defect seen in about 10% of the egg chambers from chic¹³²⁰/chic²²¹ females was a failure of proper border cell migration. As described above, profilin is highly expressed in the somatically derived border cells and allelic combinations that reduced the somatic expression of profilin delayed the migration of these cells through the nurse cell cluster (data not shown).

The one homozygous viable mutant obtained in the excision screen, chic³⁷, contained a genomic deletion between the first and second exons of chickadee. Examination of gonads from chic³⁷ homozygous animals revealed dramatic defects in the proliferation of both the male and female germline. Mutant ovaries dissected from pupae or newly eclosed adult females contained only 10 – 20 egg chambers, corresponding to one or two egg chambers per ovariole (Fig. 6B). The germarium appeared devoid of germline material, resulting in ‘empty’ follicle cell stacks (small arrow in Fig. 6B). The few egg chambers that did develop (long arrow in Fig. 6B) resembled the most severe mutant egg chambers seen in the allelic combination chic¹³²⁰/chic²²¹ described above. These egg chambers must be resorbed into the abdomen since in older females no germline tissue was visible, and the follicle cell stacks were elongated (Fig. 6C). It thus appeared that the general organization of the follicle cells was unaffected and they formed empty pseudo-ovarioles devoid of germline cells.

Testes from chic³⁷ males had a similar defect to that seen in females. Testes from newly eclosed males contained a few spermatid bundles (data not shown). Testes from older males were markedly smaller than wild type and appeared agametic (Fig. 6E). As indicated in Table 1, some less severe chic alleles were also male sterile. These alleles had a slightly milder defect, in which mutant testes still contained some spermatid bundles, but mature sperm were non-motile (data not shown; Castrillon et al., 1993).

In chic³⁷ homozygous flies, in addition to defects in oogenesis and spermatogenesis, the formation of the peripheral nervous system is disrupted. Both macrochaete and microchaete bristles on the head, thorax, legs and wings are affected to varying degrees (Fig. 7). The bristle shaft is formed as a cytoplasmic extension of the trichogen cell. Its structure is provided by a core of microtubules surrounded by fiber bundles (Overton, 1967). Staining of pupae with rhodamine-phalloidin established that the fibrillar structures were actin filament bundles (Fig. 8; Appel et al., 1993). The ridges seen on the bristle cuticle are formed by cytoplasm of the trichogen cell protruding between the actin filament bundles, and the ridges are therefore representative of the number of bundles (Fig. 7D). Normally, wild-type bristles are long and thin, and taper towards the tip (Fig. 7A, D). We compared wild-type and mutant bristles from the same region of the head bordering the eye socket. chic³⁷ mutant bristles were thicker and shorter, with sharp bends, kinks and forked ends (Fig. 7B,C,E,F,G). The ridges on mutant bristles were often thinner, more numerous and disorganized. Not every bristle on a chic³⁷ fly was affected, but almost every class of bristle had some defective members. Although there was great variety in the morphology of mutant bristles, the phenotype seen in Fig. 7C was extremely common for the ocellar bristles. Mutant bristles were also observed in hemizygous strong female sterile chickadee alleles (Table 1). In addition, large mosaic cuticle clones generated with a null allele had the same bristle morphology as chic³⁷ (data not shown).

We examined the distribution of actin filaments in bristles by staining 38-to 43-hour old pupae with rhodamine-phalloidin. In the wild-type bristles, actin was seen in 8-12 discrete bundles (Fig. 8A; Overton, 1967). Mutant bristles appeared to contain abundant actin filament bundles, but they were more numerous and somewhat thinner than wild type (Fig. 8B,C,D). Often the ends of individual bundles appeared to have separated from the core of the bristle and became bent. Overall, the actin filament bundles were more disorganized in the chic³⁷ mutant. Thus the aberrant external ridge morphology correlated with the condition of the underlying actin filament bundles.

As described above, we generated deletion mutations (chic²²¹, chic²⁸¹) that are profilin null alleles. These mutations cause a very late embryonic lethal phenotype in which the homozygous null mutant embryos develop to an advanced stage. The timing of the lethal phase is quite broad, beginning during embryonic stages 16 and 17 and ending during the first larval instar period. The few larvae that hatch from the egg cases are smaller than wild-type and display uncoordinated and greatly reduced movement (Fig. 9E). These larvae die soon after hatching. Cuticle preparations of dead embryos failed to reveal any obvious structural defects (data not shown).

To gain some insight into the role of profilin during embryogenesis, we determined the protein localization in wild-type embryos. Profilin was expressed abundantly and ubiquitously throughout embryogenesis, with specific regions of enrichment during development. During cellularization, profilin appeared to be enriched at the inner, leading edge of the cells, where actin and myosin also localize (Fig. 9A; Warn and RobertNicoud, 1990; Young et al., 1991). During gastrulation, the protein was more highly expressed in cells undergoing dramatic movements, such as those in the ventral and cephalic furrows (Fig. 9B). Staining of a stage 16 embryo revealed high levels of profilin in the ventral nerve cord (Fig. 9C).

Profilin has been found in every non-muscle cell type examined. Numerous biochemical analyses implicate profilin in the regulation of assembly of actin into filaments. Such experiments, however, have been unable to indicate whether profilin is an essential protein. We demonstrate that deletion of the chickadee gene results in a recessive embryonic lethal phenotype indicating that in Drosophila profilin is essential for proper development. The lateness of the lethal phase may be explained by a perdurance of maternally supplied profilin that is eventually degraded, inducing systemic failure in the zygotically null embryos. The fact that profilin is essential in Drosophila contrasts with yeast. Although profilin null mutations have severe effects on yeast cell growth and morphology, they are not lethal (Magdolen et al., 1988; Haarer et al., 1990). This suggests that multicellular eukaryotes have a more stringent requirement for profilin than unicellular ones.

Analysis of hypomorphic chickadee alleles reveals a phenotypic series of increasing severity, presumably corresponding to reduced transcription. Using deficiencies for the locus to halve the amount of profilin expression we find that for almost every allele, the phenotype became more severe in hemizygotes (Table 1). Our collection of chickadee alleles also reveals which steps of fly development are most sensitive to lowered profilin expression. Fertility in males and females is compromised in weaker alleles, while disruptions of bristle development and more extreme defects in the germline are found in stronger alleles. In allelic combinations that are only semiviable, defects in eye development (E. M. V., unpublished data) are added to fertility and bristle abnormalities. Finally, complete deletions for the locus are lethal. These data show that while profilin is absolutely required for viability, certain tissues are sensitive to even mild reductions in profilin expression. Similarly, in mammalian cultured cells actin filament populations have differing levels of sensitivity to overexpression of profilin (Cao et al., 1992). Taken together, these results suggest that certain populations of actin filaments are more sensitive than others to either increased or decreased profilin concentrations.

The regulation of profilin expression in Drosophila is accomplished by two transcriptional promoters in the chickadee gene (Cooley et al., 1992) rather than the presence of multiple gene copies as is the case in slime molds and Acanthamoeba (Binette et al., 1990; Pollard and Rimm, 1991). We have been unable to identify another profilin gene by low stringency blotting (unpublished data). Single copy profilin genes are also present in mice, humans and yeast (Kwiatkowski and Bruns, 1988; Magdolen et al., 1988; Sri-Widada et al., 1989). We previously determined that one of the two chickadee promoters is expressed exclusively in the ovary while the second is expressed ubiquitously (Cooley et al., 1992). Using antibody localization of profilin in egg chambers from wild-type and female sterile chickadee mutants, we now find that the ovary-specific promoter is most likely to be nurse cellspecific since nurse cell, but not follicle cell, expression is reduced in the mutant egg chambers (Fig. 2). The phenotype of this class of alleles is similarly restricted to the nurse cells where the usual cytoplasmic arrays of actin filaments fail to form.

In order to determine what function profilin is playing in cells, we have carried out a detailed analysis of the phenotypes associated with hypomorphic profilin mutations. The pleiotropic nature of these phenotypes suggests that there are roles for profilin in several distinct cellular processes, which share the need for proper regulation of a dynamic actin cytoskeleton.

The most obvious example of a role in regulating actin filament assembly was seen in mutant egg chambers that are missing the nurse cell-specific chickadee transcript (Cooley et al., 1992). Actin filament bundles that normally form in the cytoplasm fail to appear. The role of profilin in nurse cells may be to maintain the pool of actin monomers in the nurse cell cytoplasm in the ATP-bound form by stimulating nucleotide exchange. This would ensure a maximum rate of polymerization at the required time. Since filaments appear very rapidly in the cytoplasm, a mechanism must exist to trigger the dramatic actin assembly. Profilin, through its interaction with PtdIns(4,5)P₂, may be involved with transmission of an extracellular signal to trigger actin polymerization. However, the presence of PtdIns(4,5)P₂ in nurse cells has not been confirmed and we do not detect any accumulation of profilin at nurse cell plasma membranes where PtdIns(4,5)P₂ is expected to reside.

The abnormal bristles in the severe allele chic³⁷ have a very different actin phenotype than that of nurse cells. Instead of a missing population of actin filaments, actin bundles still formed in bristles, but the number of bundles was increased. The same phenotype was seen in bristles that formed within large cuticle clones of a null allele of chickadee. This result implies that high levels of profilin are not essential for the polymerization of filaments in bristle cells. We propose that in trichogen cells, profilin may function to sequester actin monomers to control nucleation of new filaments. This interpretation is based on the studies of the acrosomal reaction of echinoderm sperm (Tilney et al., 1983). In the sperm head, monomeric actin is complexed with profilin. Upon induction of the acrosomal reaction, the actin is rapidly assembled into a filamentous acrosomal process. As the filaments of the process are elongating, the profilin bound to actin monomers suppresses nucleation of new filaments, while allowing monomers to assemble onto the preferred end of established growing filaments at the same rate as free actin (Tilney et al., 1983; Pring et al., 1992). We suggest that a similar function is being played by profilin in the trichogen cell. When the expression of profilin is reduced in chic³⁷, actin nucleation becomes derepressed, leading to the formation of ectopic nuclei at the base or along the length of the bristle shaft. This results in an increased number of actin bundles.

The integrity of actin bundle organization is also compromised in chickadee. Frayed actin bundles are present in growing bristles (Fig. 8B-D) and extensively branched mature bristles are common (Fig. 7E-G). The branched bristles probably arise when cytoplasm containing stray actin bundles becomes enclosed with cuticle. The basis for actin bundle instability is not clear, although it is probably not due simply to an increase in the number of bundles. An increased number of bundles is also present in Stubble mutant bristles, but these do not undergo branching (Overton, 1967; Appel et al., 1993). We propose that the explanation is structural. The ridge pattern seen on the adult mutant bristles suggests that the underlying bundles are of varying thicknesses (Fig. 7E-G). Thinner or abnormally shaped bundles may be less structurally sound than wild-type bundles, and thus more susceptible to bending and splitting during the extension of the bristle shaft. Alternatively, an altered rate of polymerization of actin filaments caused by profilin reduction may perturb the normal interactions of filaments with bundling proteins leading to abnormal bundle assembly.

Several mutations that result in abnormal bristle formation have been characterized in Drosophila. Two of these, forked and singed, produce phenotypes especially similar to chickadee, with wavy and gnarled bristles. forked encodes proteins with no homology to any known actin binding protein (Hoover et al., 1993). However, the protein localizes to the actin bundles in wild-type developing bristles and actin bundles are not observed in forked mutant bristles (Petersen et al., 1994). This suggests that forked plays a role in actin organization. singed mutations also cause aberrant bristle morphology. The structure of the actin bundles is highly irregular in these mutants; the correct number of bundles is formed, but they are thinner and flattened (Overton, 1967). It is now known that singed encodes a protein with homology to sea urchin fascin, an actin-bundling protein (Paterson and O’Hare, 1991; Bryan et al., 1993), and that singed can bundle actin filaments in vitro (Cant et al., 1994). Thus, chickadee, singed and probably forked encode proteins required for proper actin polymerization or organization.

Dynamic actin structures play a role in the migration of cells and at least one population of migrating cells is affected in chickadee mutants. The border cells travel between nurse cells sending out cellular extensions as they migrate (Fig. 2B). Profilin is expressed abundantly in these cells and lowered doses of profilin disrupt their migration. Cellular extensions are also vital in nervous system development and the strong expression of profilin in the ventral nerve cord suggests that profilin is involved in neuron growth and extension. Such a role in neuron growth is confirmed by the recent finding that the stranded mutation (Van Vactor et al., 1993), which affects motoneuron growth cone pathfinding, is a lethal allele of chickadee (D. Van Vactor and C. Goodman, personal communication).

Egg chambers from chic¹³²⁰ hemizygous females contain a wide range of germline cell number and most of the cells are binucleate. These phenotypes provide a record of events that took place during mitotic divisions in the germarium. Normally, a germline stem cell daughter undergoes four highly synchronized divisions. Cytokinesis is not completed in each of these divisions so an interconnected cluster of 16 cells is formed. An early role of intercellular bridges is probably to facilitate the synchrony of the mitoses as has been postulated for human germline proliferation (Gondos, 1973). In chickadee mutant egg chambers, the regulation of mitoses may be disrupted such that a random number of divisions occurs. In addition, the last cytokinesis in these egg chambers appears to be wholly instead of partially incomplete resulting in predominantly binucleate cells.

These phenotypes demonstrate that profilin is required for mitosis in the germline. Its role could be to maintain a healthy actin-based cytoskeleton that can respond dynamically to signals controlling the cell cycles. When the amount of profilin falls below a certain threshold, the cells are no longer capable of supporting the normal division program. This is more starkly evident in ovaries of chic³⁷ flies in which proliferation of the germline fails entirely. The few egg chambers that form in these ovaries are probably derived from stem cells that differentiate. The egg chambers that develop have a variable number of germline-derived cells.

Profilin may also be required in specialized structures such as cleavage furrows. Studies of Tetrahymena profilin found that the protein is associated with the cytokinetic cleavage furrow (Edamatsu et al., 1992). The localization of Drosophila profilin to the leading edge of invaginating membrane during embryonic cellularization supports such a role. The mechanism of cellularization is analogous to the formation of cleavage furrows, and is mediated by actin and myosin localized at the leading edge (Warn and Robert-Nicoud, 1990; Young et al., 1991). The failure of the last cytokinesis in chickadee mutant egg chambers could correlate with exhaustion of available profilin.

Analysis of phenotypes associated with profilin mutation demonstrated that profilin is essential to normal development. Mutations that decrease profilin expression, but do not completely abolish it, result in disruption of several processes, all of which are actin dependent. The difference in actin phenotypes in germline development versus bristle development probably reflects the different requirements for actin assembly in the two systems. Further insight into profilin’s functions in vivo should be gained by genetically identifying proteins that interact with profilin.

We thank Pierre Gönczy, Steve DiNardo and Steve Wasserman for male sterile alleles of chickadee and helpful discussions; Anne Bang and Jim Posakony for help with pupal dissection; David Van Vactor and Corey Goodman for sharing and discussing their unpublished results; Ross MacIntyre and the Bloomington Drosophila Stock Center for fly stocks; the Molecular and Developmental Neurobiology Program, Yale University School of Medicine for use of the confocal microscope; Yale University School of Medicine Section of Immunology Monoclonal Antibody Facility; Mark Fortini and Barry Piekos for help with the SEM. We especially thank Pascal Goldschmidt-Clermont, Dennis McKearin, Brenda Knowles and Shalina Mahajan-Miklos for valuable comments on the manuscript and Alan Fanning and Cooley lab members for many helpful discussions. This work was supported by Public Health Service grant GM43301 (L. C.), the Pew Charitable Trusts (L. C.) and NIH Institutional Training Grant T32 HD07149 (E. M. V.).