ADF is required for cell motility during

brief communications

Cofilin/ADF is required for cell motility during Drosophila ovary development and oogenesis Jiong Chen*, Dorothea Godt†, Kris Gunsalus‡, Istvan Kiss§, Michael Goldberg‡ and Frank A. Laski*¶# *Department of Molecular, Cell and Developmental Biology and ¶Molecular Biology Institute, University of California at Los Angeles, Los Angeles, California 90095, USA †Department of Zoology, University of Toronto, Toronto M5S 3G5, Ontario, Canada ‡Department of Molecular Biology and Genetics, Cornell University, Ithaca, New York 14853, USA § Institute of Genetics, Hungarian Academy of Sciences, Szeged, H-6701, Hungary #e-mail: [email protected]

The driving force behind cell motility is the actin cytoskeleton. Filopodia and lamellipodia are formed by the polymerization and extension of actin filaments towards the cell membrane1,2. This polymerization at the barbed end of the filament is balanced by depolymerization at the pointed end, recycling the actin in a ‘treadmilling’ process2,3. One protein involved in this process is cofilin/actin-depolymerizing factor (ADF), which can depolymerize actin filaments, allowing treadmilling to occur at an accelerated rate3,4. Cofilin/ADF is an actinbinding protein that is required for actin-filament disassembly, cytokinesis and the organization of muscle actin filaments4–7. There is also evidence that cofilin/ADF enhances cell motility3,8,9, although a direct requirement in vivo has not yet been shown. Here we show that Drosophila cofilin/ADF6,10, which is encoded by the twinstar (tsr) gene, promotes cell movements during ovary development and oogenesis. During larval development, cofilin/ADF is required for the cell rearrangement needed for formation of terminal filaments, stacks of somatic cells that are important for the initiation of ovarioles. It is also required for the migration of border cells during oogenesis. These results show that cofilin/ADF is an important regulator of actin-based cell motility during Drosophila development.

useful system for the study of morphogenesis is the Drosophila ovary, which is comprised of ~20 ovarioles, each of which functions as a production line for egg development11,12. Attached to the anterior (distal) end of each ovariole, where oogenesis initiates, is a terminal filament, a stack of eight or nine flattened somatic cells that regulate the division of the adjacent germline stem cells and follicle precursor cells13,14. In addition, terminal filaments are essential for early ovary morphogenesis, as their formation is the first step in the differentiation of the ovarioles, and lack of their formation prevents ovariole development15,16. Terminal filaments are formed by cell intercalation in a process that is similar to convergent extension forms of cell rearrangement15 (Fig. 1a). Previously, we showed that formation depends on the gene bric à brac (bab), which is expressed in terminal-filament cells15,16. The analysis of bab has been instrumental in understanding filament formation, and bab encodes a putative transcriptional regulator containing a BTB/POZ domain17. Therefore, the study of bab does not directly address the molecular mechanisms involved in cell motility. To study this issue further we screened for other genes that are required for ovary development.

A

204

We studied the phenotypes and gene-expression patterns (using a lacZ reporter) of over 1,800 P-element enhancer trap lines18 in larval ovaries. One line analysed contains a pupal lethal mutation on its second chromosome. We found that this mutation affects a gene that is required for terminal-filament formation and border-cell migration (see below). This mutation maps to chromosomal region 60A7–60B6 (see Methods). The gene tsr, which encodes the Drosophila homologue of cofilin/ADF, also maps to this region6,10. Complementation analysis with known tsr alleles showed that the new mutation is a loss-of-function allele of tsr (Table 1). This was confirmed by the rescue of all phenotypic traits caused by the new mutation with a P-element construct carrying a 6.4-kilobase (kb) genomic fragment containing the wild-type tsr gene (Table 1). Cloning and sequence analysis showed that the new tsr allele, tsrntf (ntf for ‘no terminal filaments’) is not caused by a Pelement mutation, but rather has an insertion of a 7.4-kb gypsy retrotransposon19 259 base pairs (bp) downstream of the 5′ end of the first intron of tsr (Fig. 1b and Methods). Northern analysis using a tsr probe on total larval poly(A)+ messenger RNA showed that the tsrntf allele encodes a full-length transcript that is present at ~30–40% of the amount found in wild-type larvae (data not shown). The normal size of this transcript indicates that the gypsy sequences are removed from the pre-mRNA during splicing of the first intron of tsr. Descriptions of the phenotypes of different tsr alleles and transallelic combinations are summarized in Table 1. Homozygous tsrntf mutations are lethal at the pupal stage, although a few adult escapers are seen. These escapers are male sterile, and have a gnarled bristle phenotype that is similar to that caused by mutations in actin cytoskeletal proteins such as profilin, singed (a fascin homologue) and actin-capping protein (refs 20–22 and data not shown). The tsrntf-mutant ovary phenotype is characterized by a failure of terminal-filament formation at the third larval instar. Flies homozygous for the null allele tsr∆96 die as first-instar larvae, before formation of terminal filaments. The weaker loss-of-function alleles tsr1 and tsr2 are pupal lethal (tsr1) or semi-lethal (tsr2), but terminal filaments develop normally. Trans-allelic combinations between the alleles tsr1, tsr2 and tsrntf also do not cause defects in terminal-filament development. Therefore, tsrntf seems to have an allele-specific effect on the formation of terminal filaments. The allele tsr1 is caused by insertion of a P element into the 5′ untranslated region of the gene, whereas tsrntf is caused by insertion of a gypsy retrotransposon in the first intron. A possible explanation of the allele specificity is that the two insertions have different effects on the relative levels of tsr expression in various tissues, giving rise to differences in the mutant phenotypes. NATURE CELL BIOLOGY VOL 3 FEBRUARY 2001 http://cellbio.nature.com

© 2001 Macmillan Magazines Ltd

brief communications Table 1 Phenotypic comparison of tsr mutants

tsr allele 1

tsr

tsr2

tsr1

tsr2

tsrntf

• Terminal filaments present

–

–

• Pupal-lethal

• Semi-lethal at 25 ˚C, 21 males eclose

–


but quickly die (13.8% semi-survival rate,

• Pupal-lethal

n = 458*); 27.7% survival rate at 18 ˚C, only males survive and appear healthy (n = 411*), but are sterile. • Terminal filaments present

tsrntf

• Semi-viable at 25 ˚C, 21% survival

• Viable at 25 ˚C; (n = 192*)

rate (n = 114*); only males survive

• All females sterile (31 tested),

at 25 ˚C (only males survive, n = 3,069*);

and are sterile (8 tested).

some males fertile (51 tested)

2.8% survival rate at 18 ˚C (n = 2,067*);

• Viable at 18 ˚C (n = 365*)


some males fertile (44 tested);

• Semi-lethal — 0.4% survival rate

lethality occurs at pupal stage • Sterile at both 25 ˚C and 18 ˚C

females sterile (65 tested); defective

• Terminal filaments do not form at 25 ˚C,

border-cell migration

but form at 18 ˚C


• Bristle defect more severe at 25 ˚C than at 18 ˚C

tsr mutants were tested for viability, fertility and terminal-filament formation in the larval ovary and for border-cell migration in the adult ovary. Unless otherwise stated, all data were collected from animals raised at 25 °C. tsr1, tsr2 and tsr∆96 were balanced by TSTL (ref. 6), tsrntf was balanced by CyO, y+. Homozygous and transheterozygous progeny from a cross between any two of the alleles above were identified by the lack of dominant markers carried on the balancer chromosomes. * n = total number of progeny, of which one-third are expected to be tsr mutants if the allelic combination is fully viable (100% survival rate). Animals homozygous for the tsr null allele (tsr∆96/tsr∆96 ) die at first larval instar; tsr∆96 in combination with any of the other mutations leads to death at second or third larval instar, before formation of terminal filaments. P[mini-w +; tsr +] is a third-chromosome insertion of a P element carrying the tsr+ gene on a 6.4-kb genomic fragment6. Presence of P[mini-w+; tsr+] is sufficient to rescue completely all of the homozygous-mutant phenotypes.

Table 2 Migration of border cells is affecetd in tsr-mutant egg chambers Genotype

Total number of stage-10 egg chambers evaluated

Percentage of stage-10 egg chambers in which border cells have not reached the oocyte

+/tsr1 (wild-type, shifted to 25 ˚C)

100

0

tsrntf/tsr1 (maintained at 18 ˚C)

93

12

tsrntf/tsr1 (shifted to 21 ˚C)

63

71

tsrntf/tsr1 (shifted to 25 ˚C)

24

92

+/tsr1 and tsrntf/tsr1 flies were raised at 18 ˚C to adulthood. +/tsr1 adult females were shifted to 25 ˚C, and tsrntf/tsr1 adult females were either kept at 18 ˚C or shifted to 21 ˚C or 25 ˚C for 2 days before preparation of ovaries. The position of the border-cell cluster was analysed at stage 10 of oogenesis when wild-type border cells have finished their migration by reaching the oocyte.

To determine the function of cofilin/ADF during ovarian development, we analysed the tsrntf mutant phenotype using several celltype-specific markers. Staining larval ovaries with an anti-Bab antibody, which specifically identifies terminal-filament cells, showed that defined terminal-filament structures are not present in tsrntfmutant ovaries (Fig. 2c). However, Bab-expressing cells are located at roughly the correct position, and these cells also express Engrailed, another marker for terminal-filament cells (data not shown). This indicates that the tsrntf mutant phenotype is not the result of a defect in terminal-filament cell specification, but in the ability of these cells to form terminal filaments (that is, terminal-filament cells are present but there is no cell intercalation). Surprisingly, this phenotype has temperature-sensitive characteristics. In tsrntf-mutant ovaries from larvae grown at 25 °C, terminal filaments did not form (Fig. 2c), whereas at 18 °C normal filaments developed (Table 1 and Supplementary Information). At an intermediate temperature (21 °C) an intermediate phenotype was seen (Fig. 2b).

Given that the tsrntf allele is produced by a gypsy insertion into an intron, it was surprising to find that the mutation has temperaturesensitive characteristics. Temperature sensitivity is also seen in flies homozygous for the tsr2 allele (Table 1). At 25 °C homozygous tsr2 flies have a semi-lethal phenotype; some male flies eclose but quickly die. If raised at 18 °C, however, these male flies are healthy. A possible explanation for this temperature sensitivity is that reduced levels of cofilin/ADF can function adequately when development is slow (as it is at low temperature) but not when speeded up (at high temperature). It is also possible that when reduced levels of cofilin/ADF are present, the activity of another protein becomes limiting, and this protein may function better at a lower temperature. An analogous situation is seen with mutations in the yeast Sac6 gene23, which encodes fimbrin, an actin-filament-bundling protein. Mutants that lack fimbrin do not make normal actin structures and are defective in morphogenesis. Sac6 mutants are temperature-sensitive; null mutants are viable at 23 °C but have a lethal phenotype at 37 °C (ref. 23).

NATURE CELL BIOLOGY VOL 3 FEBRUARY 2001 http://cellbio.nature.com


205

brief communications a

Terminal-filament formation

Cell intercalation and sorting out of cells

b

LTR

LTR Gypsy (7.4 kb)

tsr 1

tsr ntf

1 kb

tsr 2 Exon Intron

Figure 1 Diagram of terminal-filament formation and structure of the twinstar gene. a, Three terminal filaments are shown, forming from a uniform population of terminal-filament cells during the third larval instar. The movement of the terminal-filament cells resembles a morphogenetic process known as convergence and extension, in which a group of cells flatten, intercalate horizontally and extend vertically to form a stack of cells. In addition to this, cell sorting is used to form multiple stacks. b, Molecular map of the twinstar (tsr) gene and its alleles. The 6.4-kb genomic fragment shown was used to rescue all of the tsr mutants (Table 1). tsr consists of four exons and three introns. A 7.4-kb gypsy retrotransposon inserted 259 bp downstream of the first exon–intron boundary is found in the tsrntf allele. The orientation of long terminal repeats (LTRs) of the gypsy is shown. tsr1 and tsr2 are two P-element-insertion mutations in which the P elements are inserted in the 5′ untranslated region of the first exon; their insertions sites are 19 bp apart6.

The tsrntf mutation specifically affects the shape of the terminalfilament cells and the apical cells, cells that are normally morphogenetically active during ovary development (during the pupal stage the apical cells will migrate to the posterior, forming the epithelial sheath that covers each ovariole11). In contrast to wildtype ovaries, in which terminal-filament and apical cells have a spindle-like shape that is typical of motile cells, tsrntf-mutant cells have a rounded appearance (Fig. 2c). However, the germline cells of the tsrntf mutant ovary appear normal at this stage, on the basis of their size, shape and number, and as determined by expression of the germline marker vasa (data not shown). This indicates that cofilin/ADF may directly regulate the behaviour of the mophogenetically active terminal-filament and apical cells, including changes in cell shape and motility. This is supported by the fact that a defect in the apical cell migration is seen in the tsrntf mutant (J.C., unpublished observations). The tsrntf-mutant ovary phenotype also includes a reduction in the number of terminal-filament and apical cells. It has previously been shown that cofilin/ADF is required for cytokinesis6, raising the possibility that a defect in cell division may contribute to the reduced number and increased size of the somatic cells in the tsrntf-mutant ovary. Cofilin/ADF functions by destabilizing filamentous actin3, so an increase in levels of filamentous actin would be expected in cells with reduced cofilin/ADF activity. This was previously observed in yeast cells when a temperature-sensitive allele of cofilin was inactivated by shifting the temperature to the non-permissive condition5, 206

as well as in tsr-mutant Drosophila spermatocytes6. To determine whether the apparent specificity of the tsrntf mutation for terminalfilament cells, as opposed to germline cells, reflects the effects of this mutation on filamentous actin in particular cells, we viewed filamentous actin by staining ovaries with rhodamine-conjugated phalloidin. We observed a large increase in the amount of filamentous actin in the tsrntf-mutant flies grown at 25 °C compared with wild-type flies (Fig. 3a, b); this increase was greatest in apical and terminal-filament cells. In tsrntf flies grown at 18 °C, a temperature at which terminal filaments develop normally, a much smaller increase in levels of filamentous actin was seen (see Supplementary Information). Also as expected, no increase in levels of filamentous actin was seen in ovaries from tsr2 mutants, in which the ovaries develop normally (see Supplementary Information). In addition to the increase in the filamentous-actin levels, apical and terminal-filament cells in tsrntf mutants showed an abnormal distribution of filamentous actin throughout the cytoplasm, in contrast to wild-type cells, in which filamentous actin was concentrated in the cortical cytoplasm (Fig. 3a, c). Germline cells also showed an increase in levels of filamentous actin, but it remained predominantly cortical. The strong accumulation of filamentous actin in terminal-filament and apical cells that express reduced levels of cofilin/ADF is likely to reflect a high turnover rate of filamentous actin in those cells in a wild-type ovary. In motile cells, such as terminal-filament cells, the depolymerization activity of proteins such as cofilin/ADF seems to be particularly important in counteracting the rapid polymerization of actin at the leading edge. To test further the hypothesis that cofilin/ADF is required for actin-based cell motility, we studied the effect of tsr mutations on the migration of the border cells during Drosophila oogenesis. Border-cell migration is a well established model system for studying cell migration in Drosophila24. At early stage 9 of oogenesis (Drosophila oogenesis is divided into 14 stages11,12), a wild-type egg chamber consists of an oocyte and 15 nurse cells, surrounded by an epithelial monolayer of ~650 somatic follicle cells25. During stage 9, the follicle cells are rearranged so that 95% of the cells move towards the posterior and form a columnar layer surrounding the oocyte, while the remaining follicle cells stretch to cover the 15 nurse cells (Fig. 4a–d). During this period, a group of six to ten follicle cells segregates from the follicular epithelium at the anterior pole and migrates between the nurse cells through the centre of the egg chamber towards the oocyte11,12. Late in stage 9, after migrating about 150 µm, these follicle cells reach the border between the posterior nurse cells and oocyte and are hence named border cells11. Several genes have been shown to regulate border-cell migration (reviewed in ref. 24), including the Drosophila CCAAT/enhancerbinding protein (DC/EBP), which is encoded by slow border cells and its downstream target the Drosophila FGF receptor, which is a receptor tyrosine kinase. Recently, the Rho-family GTPase Rac, the non-muscle myosin II, and the adhesion molecule DE-cadherin26 have also been shown to have essential functions in border-cell migration24. To study the potential function of cofilin/ADF in border-cell migration, we analysed tsrntf/tsr1 flies, which are viable when grown at 18 °C (they have a semi-lethal phenotype at 25 °C, most dying at the pupal stage; Table 1). Although tsrntf/tsr1 adult female flies are sterile, oogenesis usually proceeds relatively normally at 18 °C (data not shown). However, a defect in border-cell migration was seen in 12% of stage-10 egg chambers (Table 2). A stronger phenotype is seen if the adult flies are shifted to a higher temperature. Figure 4e–h, i–j shows egg chambers from tsrntf/tsr1 flies grown at 18 °C, then shifted for 2 days to 25 °C or 21 °C, respectively. In flies shifted to 25 °C border-cell migration was significantly slowed, and border cells were still in the middle of nurse cells at stage 10 and later. In 92% of mutant stage-10 egg chambers examined, border cells failed to reach the nurse cell/oocyte border. In addition to the defect in border-cell migration, some mutant egg chambers also exhibited ‘dumping’ of NATURE CELL BIOLOGY VOL 3 FEBRUARY 2001 http://cellbio.nature.com



Figure 2 Formation of terminal filaments is disrupted in tsrntf mutant ovaries. Ovaries from late-third-instar larvae are stained with anti-Bab antibody, which specifically labels terminal-filament cells. a, A tsrntf/+ ovary has a wild-type phenotype, with normal terminal filaments and apical cells (asterisk). b, At 21 °C, terminal filaments form in a tsrntf/tsrntf ovary but are not as organized and do not

contain as many cells as those in the wild-type or 18 °C tsrntf/tsrntf ovary (data not shown). c, In a 25 °C tsrntf/tsrntf ovary, terminal-filament cells express Bab protein but fail to form filaments. In addition, terminal-filament and apical cells appear larger and rounder, and the anterior region of the ovary appears elongated, relative to the wild type.

Figure 3 High levels of filamentous actin in a tsrntf mutant ovary. Confocal images of rhodamine–phalloidin-stained ovaries at the third larval instar. a, The distribution of actin filaments reveals the cell layers in a wild-type larval ovary: apical cell layer (AC), terminal-filament layer (arrow points to a terminal filament), germ-cell layer (asterisk) and basal stalk precursor cell layer (BC). b, c, A tsrntf/tsrntf ovary

has significantly higher levels of filamentous actin in apical and terminal-filament cells (bracketed) and slightly elevated levels in the germ line (asterisk). Images in a and b were obtained using identical settings; c is an image of the same ovary as b, taken at reduced contrast and brightness.

nurse-cell nuclei into the oocyte (Fig. 4g). In wild type the nursecell nuclei remain in the nurse cells until the end of oogenesis, when they break down as the contents of the nurse cells are deposited into the oocyte12. In contrast, nurse-cell nuclei were occasionally found in the oocyte at stage 10 or 11 in tsr-mutant egg chambers, and the yellowish oocyte yolk granules were found in the nurse cells (Fig. 4g). tsrntf/tsr1 females shifted to 21 °C exhibited a less severe phenotype than those shifted to 25 °C (Table 2, Fig. 4i–j). Of the stage-10 egg chambers, 71% exhibited a delay in border-cell migration, but this was the only defect observed in these egg chambers. These results indicate that cofilin/ADF is directly required for border-cell migration and that the migration defect is not a secondary response to a separate egg-chamber defect.

We have demonstrated a requirement for cofilin/ADF in cell movements during ovary development and oogenesis. During ovary development, cells expressing terminal-filament cell markers are present in tsrntf-mutant ovaries, so the defect in filament formation does not seem to be one of cell specification or differentiation, but rather in the ability of terminal-filament cells to intercalate and form stacks. The assumption that this mutant phenotype is due to defects in cell motility is supported by the presence of unusually high levels of filamentous actin in the mutant ovary, which is consistent with a loss of cofilin/ADF, and which could alter the actin cytoskeleton sufficiently to disrupt cell movement. We have also shown that cofilin/ADF is required for migration of border cells during oogenesis. By extension, we proposed that tsr may be required for many, perhaps all, actin-based cell movements in



207


Figure 4 Migration of border cells in wild-type and tsr-mutant egg chambers. The tsr1 enhancer trap line expresses β-galactosidase at high levels in the nuclei of border cells and nurse cells, and at lower levels in the nuclei of the oocyte and some of the follicle cells. a–d, In phenotypically wild-type tsr1/+ egg chambers, border cells (arrowheads) migrate from the anterior pole of the egg chamber between the nurse cells during stage 9 (a, b), and by early stage 10 (c, d) have

already reached the surface of the posteriorly located oocyte. e–j, In egg chambers from tsrntf/tsr1 females raised at 18 °C and then shifted to 25 °C (e–h) or 21 °C (i, j) for 2 days, border cells exhibit a migration defect. These border cells often fail to reach the oocyte by stage 10 (e–g, i, j) and are sometimes still in the middle of nurse cells by stage 11 (h). c, e and f show stage-10A egg chambers; d, g, i and j show stage-10B egg chambers.

Drosophila. Understanding how cofilin/ADF is coordinately regulated to allow formation of terminal filaments and migration of border cells should provide insight to its function in many developmental systems. Recently, a mammalian cultured cell system was used to show that LIM-kinase 1 (LIMK-1) phosphorylates and inactivates cofilin/ADF, and that the GTPase Rac is capable of regulating the activity of LIMK-1 (refs 27, 28). Further study of systems that require cofilin/ADF for normal development may allow the determination of how these signal-transduction pathways control the dynamic alterations of the actin cytoskeleton in developing animals.

second intron (at positions 4,581 and 4,627 of RS6.4; ref. 6), neither of which are in the splicing consensus sites of the intron. Wild-type and tsrntf/tsrntf mRNAs from third-instar larvae were isolated using the Direct mRNA kit (Qiagen). RNA blotting was carry out using standard methods. Membranes were hybridized with a 32Plabelled tsr probe made from tsr complentary DNA6 and with a ribosomal protein 49 (rp49) DNA probe. The relative expression level of tsr mRNA was estimated using a phospho-imager with the rp49 transcript as a control.

Methods Genetics. The tsrntf mutation was found in fly line l(2)k08106 (ref. 18), which contains five independent P-element insertions29, although none of these insertions caused the tsrntf mutation, as a line in which all the insertions were removed by recombination continued to exhibit the tsrntf mutant phenotype. The stock y w; tsrntf/CyO,y+ was generated by recombining away the five P-element insertions in the l(2)k08106 line. Meiotic recombinant mapping, using a multiply marked second chromosome (al dp pr b c px sp; Bloomington Stock Center, Bloomington, IN), placed tsrntf between px (58F) and sp (60B13-C4). The deficiency chromosomes Df(2R)orBR-11 (59F6-8; 60A8-16) and Df(2R)265lex36,30 (60A7-12; 60B3-6) failed to complement tsrntf, placing tsrntf between polytene bands 60A7 and 60B6. Alleles of tsr, which lies in the region of 60A10–B1, did not complement tsrntf (Table 1). To carry out complementation tests, tsr1, tsr2 and the null allele tsr∆96 (ref. 6), which are balanced over TSTL, were each crossed to tsrntf balanced by the second-chromosome balancer CyO,y+. The trans-heterozygous progeny from a cross between tsrntf and any of the tsr alleles were scored for the absence of Cy, which is carried on both of the balancer chromosomes. A third chromosome insertion of the P-element construct P[mini-w+; tsr+] (ref. 6) that carries a 6.4-kb wild-type genomic fragment of the tsr gene completely rescued the ovary phenotype and pupal lethality of tsrntf (Table 1). To carry out the rescue experiment, y w/Y; tsrntf /CyO; P[mini-w+; tsr+]/+ flies were crossed with y w/y w; tsrntf/CyO,y+; + flies. The y w; tsrntf/tsrntf; P[mini-w+; tsr+]/+ progeny were viable, apparently healthy and were fertile. To determine the presence of germline cells in tsrntf ovaries, the P-element enhancer trap line BC69, which expresses β-galactosidase in germ cells, was crossed into a tsrntf/tsrntf background. BC69 (J. L. Couderc and F.A.L., unpublished observations) has a P-lacZ element inserted in the vasa gene (35C1 on the second chromosome).

Molecular analysis of tsrntf. Genomic DNA from wild-type (Canton S) and tsrntf-homozygous animals at third larval instar were extracted using phenol chloroform in accordance with standard methods. Regions of tsr including all four exons and three introns were amplified by polymerase chain reaction (PCR; Elongase from GibcoBRL) from the wild-type and tsrntf/tsrntf genomic DNA. Sequence analysis of a 9.8-kb tsrntf/tsrntf PCR product (PCR of wild-type DNA produced a 2.4-kb band) revealed a 7.4-kb gypsy element19 inserted 259 bp downstream of the first exon–intron boundary. The orientation of this gypsy is such that its two long terminal repeats (LTRs) were in the opposite direction to that of the tsr transcript. No mutations were identified within the coding region of tsrntf, but two C→A mutations were found in the

208

Straining and microscopy. Parental y w; tsrntf/CyO,y+ yielded female progeny of the genotype y w; tsrntf/tsrntf. These larvae were isolated on the basis of their yellow phenotype. Both wild-type and tsrntf-mutant female larvae were aged and selected at the late third larval instar/puparium-formation stage as described15. Their ovaries were dissected and stained with anti-Bab-R2 antibody or with rhodamine–phalloidin (Molecular Probes) as described15. The adult ovaries of tsr1/+ and tsrntf/tsr1 animals were dissected in PBS and stained with Xgal as described15. Stained ovaries and egg chambers were mounted in 50% glycerol for differential image-contrast microscopy and in Vectashield mounting medium (Vector Laboratories) for confocal microscopy. RECEIVED 8 DECEMBER 1999; REVISED 7 AUGUST 2000; ACCEPTED 9 OCTOBER 2000; PUBLISHED 18 JANUARY 2001.

1. Mitchison, T. J. & Cramer, L. P. Cell 84, 371–379 (1996). 2. Carlier, M. F. Curr. Opin. Cell Biol. 10, 45–51 (1998). 3. Carlier, M. F. et al. J. Cell Biol. 136, 1307–1322 (1997). 4. Bamburg, J. R. Annu. Rev. Cell Dev. Biol. 15, 185–230 (1999). 5. Lappalainen, P. & Drubin, D. G. Nature 338, 78–82 (1997). 6. Gunsalus, K. C. et al. J. Cell Biol. 131, 1243–1259 (1995). 7. Ono, S., Baillie, D. L. & Benian, G. M. J. Cell Biol. 145, 491–502 (1999). 8. Aizawa, H., Sutoh, K. & Yahara, I. J. Cell Biol. 132, 335–344 (1996). 9. Loisel, T. P., Boujemaa, R., Pantaloni, D. & Carlier, M. F. Nature 401, 613–616 (1999). 10. Edwards, K. A. et al. Proc. Natl Acad. Sci. USA 91, 4589–4593 (1994). 11. King, R. C. Ovarian Development in Drosophila melanogaster (Academic Press, New York, 1970). 12. Spradling, A. in Development of Drosophila melanogaster (eds Bate, M. & Martinez-Arias, A.) 1–70 (Cold Spring Harbor Press, Cold Spring Harbor, New York, 1993). 13. Lin, H. & Spradling, A. Dev. Biol. 159, 140–152 (1993). 14. Forbes, A., Lin, H., Ingham, P. & Spradling, A. Development 122, 1125–1135 (1996). 15. Godt, D. & Laski, F. A. Development 121, 173–187 (1995). 16. Sahut-Barnola, I., Godt, D., Laski, F. A. & Couderc, J. L. Dev. Biol. 170, 127–135 (1995). 17. Zollman, S., Godt, D., Prive, G. G., Couderc, J. L. & Laski, F. A. Proc. Natl Acad. Sci. USA 91, 10717–10721 (1994). 18. Torok, T., Tick, G., Alvarado, M. & Kiss, I. Genetics 135, 71–80 (1993). 19. Marlor, R. L., Parkhurst, S. M. & Corces, V. G. Mol. Cell. Biol. 6, 1129–1134 (1986). 20. Verheyen, E. & Cooley, L. Development 120, 717–728 (1994). 21. Cant, K., Knowles, B., Mooseker, M. & Cooley, L. J. Cell Biol. 125, 369–380 (1994). 22. Hopmann, R., Cooper, J. A. & Miller, K. G. J. Cell Biol. 133, 1293–1305 (1996). 23. Adams, A. E., Botstein, D. & Drubin, D. G. Nature 354, 404–408 (1991). 24. Montell, D. J. Cell Biochem. Biophys. 31, 219–229 (1999). 25. Margolis, J. & Spradling, A. Development 121, 3797–3807 (1995). 26. Niewiadomska, P., Godt, D. & Tepass, U. J. Cell Biol. 144, 533–547 (1999). 27. Yang, N. et al. Nature 393, 809–812 (1998).



brief communications 28. Arber, S. et al. Nature 393, 805–809 (1998). 29. Spradling, A. et al. Proc. Natl Acad. Sci. USA 92, 10824–10830 (1995). 30. Reed, B. H. & Orr-Weaver, T. L. Development 124, 3543–3553 (1997).

ACKNOWLEDGMENTS We thank H. D. Pham for excellent technical assistance, J. Pendelton, J. Peredo and T. Tran for their assistance in screening the 1,800 P-element lethal lines, and M. Bejar, J. Monroy and L. Ng for help with

mapping tsrntf. We also thank T. Orr-Weaver, P. Morcillo, the Bloomington Stock Center and T. Laverty (Berkeley Drosophila Genome Project) for stocks, and U. Tepass for critical reading of the manuscript. This work was supported by USPHS National Research Service Awards GM07185 and GM07617 (to J.C. and K.G., respectively). D.G. was supported by the Natural Science and Engineering Research Council of Canada. F.A.L. and M.G. were supported by NIH grants GM40451 and GM48430, respectively. Correspondence and requests for materials should be addressed to F.A.L. Supplementary Information is available on Nature Cell Biology’s website (http://cellbio.nature.com) or as paper copy from the London editorial office of Nature Cell Biology.



209

supplementary information

Figure S1 Third-larval-instar ovaries grown at different temperatures. Confocal images showing rhodamine–phalloidin-stained ovaries grown at 25 °C (a–c), or 18 °C (d–f). Ovaries were dissected from phenotypically wild-type tsrntf/+ (a, d), tsr2/tsr2 (b, e) and tsrntf/tsrntf (c, f) larvae. At both 18 °C and 25 °C, levels of filamentous actin in tsr2 ovaries (b, e) are similar to those in the wild type (a, d), and normal terminal filaments are present. tsrntf ovaries grown at 25 °C do not con-

tain terminal filaments and have much higher levels of filamentous actin than wildtype ovaries (compare a and c). tsrntf ovaries grown at 18 °C (f) have terminal filaments and significantly less filamentous actin than those grown at 25 °C (c), although their levels of filamentous actin are slightly higher than those of the wild type (compare d and f).



1