Journal of Science 107, 1863-1873 (1994) 1863 Printed in Great Britain © The Company of Biologists Limited 1994

The nullo protein is a component of the actin-myosin network that mediates cellularization in embryos

Marya A. Postner and Eric F. Wieschaus* Department of Molecular Biology, Princeton University, Princeton, NJ 08544, USA *Author for correspondence

SUMMARY

After the 13th nuclear division cycle of Drosophila embryo- actin caps and within metaphase furrows. In cellularizing genesis, cortical microfilaments are reorganized into a embryos, nullo co-localizes with the actin-myosin network hexagonal network that drives the subsequent cellulariza- and invaginates along with the leading edge of the plasma tion of the syncytial embryo. Zygotic transcription of the membrane. The serendipity-α (sry-α) protein co-localizes nullo and serendipity-α genes is required for normal struc- with nullo protein to the hexagonal network but, unlike the turing of the microfilament network. When either gene is nullo protein, it localizes to the sides rather than the deleted, the network assumes an irregular configuration vertices of each hexagon. Mutant embryos demonstrate leading to the formation of multinuceate cells. To investi- that neither protein translationally regulates the other, but gate the role of these genes during cellularization, we have the localization of the sry-α protein to the hexagonal made monoclonal antibodies to both proteins. The nullo network is dependent upon nullo. protein is present from cycle 13 through the end of cellu- larization. During cycle 13, it localizes between interphase Key words: Drosophila embryo, cytokinesis, contractile ring

INTRODUCTION taposition: the sides of each polygon in the array are formed by converting the ‘fuzzy’ actin organization at the cap margins In Drosophila embryos, the early nuclear divisions are not into more finely aligned actin filaments (Simpson and followed by cytokinesis and the embryo initially develops as a Wieschaus, 1990). Each hexagonal interface of the actin . This organization persists until after the 13th network defines the site of membrane invagination. Cytoplas- division, at which time the embryo consists of approximately mic myosin co-localizes with actin in the hexagonal network 6,000 nuclei arranged in a monolayer in the embryo’s cortex. (Warn et al., 1980; Young et al., 1991), and both filamentous Subdivision of the cortical cytoplasm into individual cells is actin and functional cytoplasmic myosin are required for known as cellularization. During this process, plasma membrane invagination (Zalokar and Erk, 1976; Foe and membrane invaginates from the surface in a roughly hexagonal Alberts, 1983; Kiehart et al., 1990). Contraction of this pattern, precisely separating each nucleus from its immediate actin/myosin array has been postulated to provide a mechanis- neighbors. Once the membrane has reached a depth of about tic force driving the invagination. This role for the actin- 25 µm, the base of the invaginating membrane furrow begins myosin network is based in part on an analogy with the ‘con- to widen, eventually separating the newly formed cells from tractile rings’ of actin and myosin that are thought to drive the underlying yolk. The resulting cellular blastoderm consists invagination of the plasma membrane during conventional of a single layer of columnar cells surrounding the central yolk cytokinesis (Mabuchi, 1986; Salmon, 1989; Schroeder, 1990; sac. Satterwhite and Pollard, 1992). During the first ten minutes of cycle 14, a highly organized Most of the components of the hexagonal array are supplied array of F-actin is formed on the cytoplasmic face of the by maternal transcription during oogenesis and are already invaginating plasma membrane (Fig. 1, see also Warn and present as RNA or protein in the unfertilized egg. Cellulariza- Magrath, 1983; Simpson and Wieschaus, 1990; Warn and tion, however, marks the point in Drosophila development Robert-Nicoud, 1990; Schejter and Wieschaus, 1993). Prior to when the embryo becomes dependent on gene products formation of the array, the cortical actin of the embryo is supplied by the embryo’s own transcription (Arking and organized in ‘caps’ overlying each nucleus. Initially the caps Parente, 1980; Edgar and Schubiger, 1986). A small number of formed in cycle 14 resemble those seen in the preceding inter- genes have been identified whose zygotic products are required phases. However, in contrast to earlier caps, which remain for the formation of a normal actin array (Wieschaus and static during interphase, the cycle 14 caps soon enlarge until Sweeton, 1988; Merrill et al., 1988). Embryos lacking either the their bases touch. The actin array arises in regions of cap jux- nullo or the serendipity-alpha (sry-α) gene show very similar 1864 M. A. Postner and E. F. Wieschaus abnormalities in the actin array: some of the sides of the cies that uncover the nullo locus. The deficiencies Df(1)6F1-2 and hexagons are unusually thick while others are extremely thin or Df(1)LIMDF were most commonly used (for description see Simpson missing altogether (Fig. 1B,C). In nullo embryos, the initial few and Wieschaus, 1990; Rose and Wieschaus, 1992). Embryos with the minutes of network formation appear normal (Wieschaus and sry-α phenotype were collected from a stock that is heterozygous for α Sweeton, 1988; Simpson and Wieschaus, 1990). However, at Df(3R)X3F, which uncovers the sry- gene. Since no point mutations the onset of membrane invagination, network formation is exist for either gene, the mutant phenotype only arises in deficiency embryos. The terms ‘nullo mutant’ and ‘sry-α mutant’ are used to incomplete and the uneven distributions and sporadic disrup- describe the deficiency embryos. tions in the actin-myosin network become obvious. Once underway, network and membrane invagination appear to Production and screening of monoclonal antibodies proceed normally: neither the kinetics of membrane extension Monoclonal antibodies to the nullo protein were generated using a nor the length of the newly formed cells is significantly different nullo-glutathione S-transferase fusion protein as the antigen. An in- from that observed in wild-type embryos. However, cleavage frame fusion of the entire nullo protein to the carboxyl terminus of furrows do not invaginate where the network is discontinuous glutathione S-transferase (Smith and Johnson, 1988) was constructed and multinucleate cells form. The only obvious difference in the following manner. The nullo coding region was PCR amplified between nullo and sry-α mutant embryos is that sry-α embryos from the nullo M1 cDNA (Rose and Wieschaus, 1992) using primers have fewer multinucleate cells (Merrill et al., 1988; homologous to the ends of the coding region. The primers also contained an external stretch of bases that lacked homology to nullo Schweisguth et al., 1990; E. Schejter, personal communication). α and contained an EcoRI restriction site. The resulting PCR product Molecular characterization of the nullo and sry- genes has was digested with EcoRI and ligated into the EcoRI site of pGEX-2T revealed that both genes encode single, blastoderm-specific (Pharmacia). The orientation of the inserts was determined by restric- transcripts that are uniformly distributed throughout the tion mapping. syncytial embryo and accumulate in large amounts over a short The pGEX-2T-nullo plasmid was transformed into Escherichia coli period of time (Vincent et al., 1985; James and Vincent, 1986; strain JM101 and the production of fusion protein was induced with Rose and Wieschaus, 1992). Transcript levels reach a sharp 1 mM IPTG (Pharmacia). After 3 hours, the cells were harvested. peak around the onset of cellularization and subsequently Because the fusion protein was stubbornly insoluble, it was excised decrease in a rapid, spatially patterned manner. The main dif- from a preparative acrylamide gel, electroeluted with Elutrap ference in the transcription pattern of the two genes is that sry- (Schleicher and Schuell), and dialyzed against MTPBS (Smith and α Johnson, 1988). The purified protein was used to immunize two mice transcripts arise, peak and decline slightly later than nullo. and monoclonal antibodies were produced following standard Neither gene is required for the transcription of the other (Rose protocols (Harlow and Lane, 1988). Supernatants from the monoclonal and Wieschaus, 1992). lines were tested by western blot for recognition of the fusion protein The sry-α protein is 58 kDa in size, lacks extensive and of glutathione S-transferase. The 23 lines that recognized only the homology to any known proteins and shows few structural fusion protein were tested by western blot for reactivity with proteins motifs (Ibnsouda et al., 1993). Immunolocalization indicated from two- to three-hour Ore-R embryos. Monoclonal supernatants that during cellularization sry-α protein localizes to the leading 5C3-12 and 2F8-18 specifically recognize the nullo proteins. Because edge of the invaginating plasma membrane (Schweisguth et al., the 5C3-12 antibody reacts more strongly with the nullo proteins than 1990). Like its transcript, the sry-α protein is short-lived. The does 2F8-18, it was used preferentially unless otherwise indicated. Monoclonal antibodies to the sry-α protein were generated using a nullo gene is predicted to encode a 23 kDa protein lacking α α truncated version of the sry- protein as the antigen. This protein, homology to known proteins, including the sry- protein. which contains amino acids 46 to 530 of sry-α, was produced from a Sequence analysis demonstrated that the nullo protein has an T7 RNA polymerase-inducible vector (Studier and Moffat, 1986). The excess of basic amino acids (predicted pI is 11.4) and plasmid, pPαNN, contains a NarI to NcoI fragment of the sry-α gene suggested that the protein may be myristoylated (Rose and cloned in pET3a. It was generously provided by Alain Vincent. The Wieschaus, 1992). However, previous studies did not address plasmid was transformed into E. coli strain BL21-Lys S. Fusion intracellular localization of the nullo protein or its specific cell protein production was induced for three hours with 1 mM IPTG. The biological function during cellularization. cells were lysed by freezing with 3.3 mg/ml lysozyme and then While many broad questions regarding the mechanisms of thawing. Inclusion bodies were purified by isolating all proteins cellularization remain unanswered, several specific questions insoluble in DOC buffer (200 mM NaCl, 1% sodium deoxycholate, about the nullo and sry-α proteins are amenable to experi- 1% NP40, and 1 mM DTT). The protein pellet was washed three times with 0.5% Triton X-100, 1 mM EDTA and 1 mM DTT, before being mentation: do the proteins directly interact with the cellular- resuspended in TE (10 mM Tris, 1 mM EDTA). This protein prepa- ization machinery or are they indirect participants in cellular- ration was greatly enriched for the truncated sry-α protein. Two mice ization? What are the functions of the nullo and sry-α proteins were immunized with it and monoclonal antibodies were produced. during the process? Is the similarity of their mutant phenotypes The monoclonal supernatants were screened by western blot to test due to their participation in a common pathway or do the two reactivity with the truncated sry-α protein that served as the antigen. proteins function independently? To address these questions, Reactive supernatants were further tested for reactivity with a 58 kDa we have generated monoclonal antibodies to both nullo and protein from two- to three-hour wild-type embryos. Monoclonal sry-α proteins, and examined their distribution during cellu- supernatants 1G10, 3H6, 4G4, 6B12 and 6F4 all react strongly with α larization in both wild-type and mutant embryos. the sry- protein. Antibody 1G10 was used in most instances. Immunoprecipitation MATERIALS AND METHODS Two- to three-hour wild-type embryos were dechorionated, ground in 80 mM Tris with 2% SDS, and boiled for five minutes. The extracts Genotypes and stocks used were chilled on ice. Modified RIPA buffer (50 mM Tris, 300 mM Ore-R was used as the wild-type stock. Embryos with the nullo NaCl, 1% NP40, 0.5% sodium deoxycholate) and 10% Triton X-100 phenotype were collected from balanced stocks containing deficien- were added to the extracts to achieve a final concentration of 0.2%

nullo protein in Drosophila embryos 1865

SDS and 1% Triton X-100. Samples of 75 µl each of nullo mono- clonal antibodies 5C3-12 and 2F8-18 were incubated with the protein extracts for one hour at 4ûC. Then, 50 µl of 50% Protein A-Sepharose beads (Pharmacia) in RIPA buffer (above recipe plus 1% SDS) were added and incubated at 4ûC. After one hour, the pellet was recovered and washed twice in RIPA buffer. Antibody staining of embryos Two- to three-hour embryos were dechorionated and fixed by one of two procedures: (1) fixation for 20 minutes with 18.5% formaldehyde saturated with heptane followed by manual devitellinization or (2) boiling for 10 seconds in Triton-salt solution (68.4 mM NaCl, 0.03% Triton X-100) followed by the addition of a vast excess of ice-cold Triton-salt solution, devitellinization using methanol and heptane, and post-fixation of at least one hour in methanol. The fixed and devitellinized embryos were incubated for one hour at room temper- ature with PBT10 (PBS with 10% BSA and 0.1% Tween-20). Incu- bations with the primary antibodies were performed overnight at 4ûC: sry-α monoclonal antibody 1G10 was diluted 1:25 in PBT1 (PBS with 1% BSA and 0.1% Tween-20); nullo monoclonal 5C3-12 was diluted 1:15 in PBT1; antisera to cytoplasmic myosin (kindly provided by Dan Kiehart) was diluted 1:250 in PBT1. After incubation with the appropriate primary antibody, the embryos were washed once with PBT1 and four times for 30 minutes each with BNT100 (PBS with 2% normal goat serum, 100 mM NaCl, 1% BSA, and 0.1% Tween- 20). Preabsorbed fluorescent secondary antibodies were diluted 1:250 Fig. 1. The actin-myosin network that forms in Drosophila embryos in PBT0.1 (PBS with 0.1% BSA and 0.1% Tween-20) and embryos during cycle 14. In wild-type embryos (A), the array consists of were incubated with them at room temperature for one to three hours. approximately 6000 polygons of roughly equal size. Each polygon To visualize filamentous actin, embryos were stained for 20 minutes defines the area above a single somatic nucleus. Bar, 100 µm. At µ µ with either 0.165 M bodipy-phalloidin or 0.165 M rhodamine-phal- higher magnification, the wild-type array (B) shows a very regular loidin (Molecular Probes). After several washes in PBS-Triton, they configuration of polygons composed of uniformly thick actin µ were incubated for 3 minutes with 1 g/ml Hoechst 33258 (Poly- interfaces. In embryos deficient for nullo (C) or sry-α (D), the array sciences), a DNA-specific dye. The embryos were washed extensively has interfaces of irregular thickness and the individual polygons are in PBS-Triton and PBS before being mounted in Aquapolymount of variable size. Actin microfilaments were visualized by staining (Polysciences). Embryos were examined and photomicrographs made with FITC-labeled phallacidin. Bar in D (also applies to B and C), using a Bio-Rad MRC600 confocal microscope. 10 µm.

RESULTS protein remain high during the initial slow phase of cellular- nullo protein is blastoderm-specific ization, when invagination of membrane is thought to depend In order to characterize the distribution of nullo protein, mon- on the incorporation of new membrane behind the furrow oclonal antibodies to the protein were isolated. The antigen (Turner and Mahowald, 1976). Levels drop rapidly during the was a fusion protein consisting of glutathione S-transferase and subsequent ‘fast’ phase and protein is barely detectable by the the full-length nullo protein. Supernatants from two mono- beginning of . Both forms of nullo protein show clonal lines recognized proteins from two- to three-hour-old similar kinetics of accumulation and disappearance. The embryos in the size range predicted for nullo (approximately pattern parallels that previously reported for the nullo RNA 23 kDa). Both supernatants reacted with the same protein with a lag of about fifteen to twenty minutes. doublet of 25 and 26.5 kDa (Fig. 2A). The doublet was absent in protein preparations from embryos homozygous for a small Intracellular localization of nullo proteins deficiency of nullo (Fig. 2B and C), indicating that both Antibodies to nullo were used for immunolocalization of the proteins are encoded by the nullo gene and that both mono- protein during syncytial and cellular blastoderm stages. To clonal lines are specific for nullo proteins. The difference in visualize the microfilament network independently, wild-type migration between the two forms of nullo is probably due to a embryos were simultaneously stained with antibodies against post translational modification, since northern blots and myosin or with phalloidin. nullo protein is first detectable in sequence analysis predict only a single nullo product (Rose and whole-mount embryos during interphase of cycle 13 (Fig. 3A). Wieschaus, 1992). At this stage, all detectable nullo protein is localized in the The temporal profile of nullo protein levels was determined cortical cytoplasm of the embryo and is punctate or vesicular by western blot analysis of single Ore-R embryos staged prior in nature. The protein is restricted to a region apical to the to homogenization (Fig. 1D). nullo protein was detectable monolayer of nuclei, and appears to be associated with the between interphase of cycle 13 and the beginning of gastrula- plasma membrane. It does not co-localize with the actin caps tion. The amount of nullo protein is low in cycle 13 interphase that form above the interphase nuclei and instead appears to be embryos, increases greatly during the 13th division, and restricted to the areas between the caps. When viewed from the reaches an apparent peak in early cycle 14. Levels of nullo surface, the resulting pattern of nullo distribution resembles a 1866 M. A. Postner and E. F. Wieschaus

Fig. 2. Western blots of embryo extracts using nullo monoclonal antibody. (A) Extracts from 2- to 3-hour wild-type embryos. Lane 1 contains approximately 10 embryos; lane 2 contains approximately 5 embryos; and lane M contains unstained low molecular mass markers (Bio-Rad). The monoclonal antibody recognizes two proteins of approximately 25 and 26.5 kDa. (B) Each lane contains proteins extracted from a single cycle 14 embryo. The embryos were collected from a cross in which one-quarter of the embryos are deleted for the nullo locus. The embryos were randomly harvested during early cycle 14 and no attempt was made to pick normal or mutant embryos. The embryos in lanes 2, 4, 8 and 11 lack both nullo proteins and are assumed to be deleted for the nullo locus. (C) In this control experiment, each lane contains proteins extracted from a single wild-type embryo in cycle 14. As expected, two forms of the nullo protein were recovered from all embryos. (D) Wild-type embryos were carefully staged and harvested at precise stages, from the beginning of cycle 13 through early gastrulation. Each lane contains protein from a single embryo. Lanes 1 and 2, cycle 13, interphase; lanes 3 and 4, cycle 13, mitosis; lanes 5 and 6, cycle 14, pre- cellularization; lanes 7 and 8, slow phase of cellularization; lane 9, beginning of fast phase of cellularization; lanes 10 and 11, end of fast phase; and lanes 12 and 13, gastrulation. The levels of the nullo protein are highest from the 13th division through the slow phase of cellularization.

relatively regular network of interlocking rings that encom- During the next ten minutes, as the network matures, nullo passes the entire embryo (Fig. 3A). The intensity of staining protein is colinear with the network and shows an intense and with the nullo antibody is weak at this stage. orderly punctate staining pattern (Fig. 4E,F,G,H). Closer During the mitosis of cycle 13, nullo protein is localized analysis of the staining patterns within each hexagonal unit within the pseudocleavage furrows that transiently invaginate reveals that nullo and actin are distinguishable. When visual- between the dividing nuclei. Although actin underlies the ized with phalloidin or phallacidin, the hexagonal units are entire plasma membrane during this stage, nullo appears to be defined by uninterrupted ‘lines’ of filamentous actin. The ends associated only with invaginated membrane regions. When of these lines meet to form the hexagon’s vertices (Rose and viewed from above (Fig. 3B), nullo forms elongated hexagons, Wieschaus; see also Fig. 5). By comparison, nullo protein each enclosing a mitotic nucleus. The intensity of nullo shows a discontinuous, punctate staining pattern. Many dots of antibody staining at this stage is considerably greater than that nullo protein are aligned to give the hexagonal pattern. While in interphase of cycle 13. there is nullo present along the sides of each hexagon, the At the onset of cycle 14, when actin caps re-form and the highest concentration of nullo staining occurs at the vertices. hexagonal array of actin and myosin begins to resolve, nullo In addition, some nullo protein is present in the apical regions protein localizes to the bases of the caps (Fig. 4A,B,C,D). above the nuclei, as well as in the cytoplasm just below the

Fig. 3. Localization of the nullo protein in late syncytial embryos. When wild-type embryos at interphase of cycle 13 (A) are viewed from surface, the nullo protein forms a hexagonal network. The nullo protein localizes to the pseudocleavage furrows during the mitosis of cycle 13 (B). It forms a hexagonal array in cycle 14 (C). The absolute level of staining with the nullo antibody dramatically increases from interphase of cycle 13 to early cycle 14. To compare spatial patterns of distribution, all panels have been printed at the same intensity levels. Bar in C (also applies to A and B), 10 µm. nullo protein in Drosophila embryos 1867

Fig. 4. Cellularizing wild-type embryos double labelled with the nullo antibody (A,C,E,G,I,K,M,O,Q,S) and cytoplasmic myosin antibody (B,D,F,H,J,L,N,P,R,T). The first and second columns show surface views, the third and fourth columns show cross-sections from the same embryos. During conversion of the actin caps to a hexagonal network (A-D), the nullo protein is localized to the forming network (A,C). It maintains a localization to the leading edge of the invaginating membrane at the initiation of membrane invagination (E-H) and during slow phase (I-L). During the fast phase of cellularization (M-P), the levels of nullo protein present at the cellularization front decrease and the protein becomes dispersed throughout the cytoplasm. By the completion of cellularization (Q-T) nullo protein is almost undetectable. Bar in D (applies to entire figure), 10 µm. 1868 M. A. Postner and E. F. Wieschaus bases of the nuclei, where its staining pattern gives the impres- hexagonal array. In some confocal cross-sections, nullo sion of being vesicular. staining appears as a line extending apically from the furrow During the slow phase of cellularization (Fig. 4I,J,K,L), canal. This line never extends more than half-way to the apical nullo protein continues to be primarily associated with the plasma membrane. Unlike actin, nullo is not localized to the plasma membrane at and slightly apical to the invaginating non-invaginated plasma membrane on the apical surface, furrow canal, where it maintains its initial distribution in a although some slight staining above background is observed in

Fig. 5. Cellularizing wild-type embryos double labelled with the sry-α antibody (A,C,E,G,I,K,M,O) and phalloidin (B,D,F,H,J,L,N,P). The first and second columns show surface views; the third and fourth columns show cross-sections from the same embryos. During actin-myosin network resolution (A-D) and during slow phase of cellularization (E-H), sry-α protein localizes to the hexagonal array (see arrows in C and G). The levels of localized sry-α protein remain high at the end of slow phase (I-L) but decrease during the latter half of fast phase (M-P). The protein becomes more dispersed in the cytoplasm (M, O), although some staining remains in a hexagonal pattern (M). sry-a protein also localizes in a sphere above each nucleus (G and K, arrowheads) until the final stages of cellularization (G,K,) as well as maintaining a more general localization in the apical cytoplasm (G,K). Bar in D (applies to entire figure), 10 µm. nullo protein in Drosophila embryos 1869 the cytoplasm apical to the nuclei. All cytoplasmic staining previous stages of cellularization. When viewed tangentially at observed at these stages is still punctate in appearance. the level of the furrow canals, the nullo protein continues to As cellularization progresses into the fast phase (Fig. form a thin network of interconnecting hexagons. This is in 4M,N,O,P), nullo protein maintains its localization to the cel- contrast to the actin-myosin network, which thickens, resulting lularization front, but the intensity of staining is significantly in a ring-like appearance of the individual units in the network. diminished and more variable from embryo to embryo than in During the course of fast phase (Fig. 4Q,R,S,T), progressively

Fig. 6. Localization of the nullo proteins in nullo and sry-α mutant embryos. Embryos deleted for nullo locus (E,F,G,H) and their heterozygous siblings (A,B,C,D) were double labelled using nullo antibody (A,C,E,G) and cytoplasmic myosin antisera (B,D,F,H). The specificity of the nullo antibody is demonstrated by its failure to stain nullo mutant embryos (E,G). In embryos lacking sry-α, (M,O), the nullo protein forms a network that is collinear with the actin-myosin hexagonal array: a relatively normal nullo network is seen everywhere except where the actin- myosin network is disrupted. Bar in D (applies to entire figure), 10 µm. 1870 M. A. Postner and E. F. Wieschaus less nullo protein is localized to the furrow canal and the Intracellular localization of the serendipity-alpha majority of nullo protein becomes dispersed throughout the protein cytoplasm. The intensity of nullo staining continues to decline Since nullo and serendipity-alpha (sry-α) mutants have a and is no longer detectable above background when cellular- common phenotype and both proteins have now been reported ization is completed. to localize to the hexagonal array of actin, a detailed compar-

Fig. 7. Localization of the sry-α protein in nullo and sry-α mutant embryos. Comparison of embryos deleted for sry-α (E,F,G,H) and their heterozygous siblings (A,B,C,D) when double labelled with sry-α antibody (A,C,E,G) and phalloidin (B,D,F,H) confirm that the sry-α antibody is specific: sry-α heterozygotes have a hexagonal array of sry-α protein (A,C) whereas their mutant siblings show no such staining (E,G). Embryos deleted for nullo (M,N,O,P) and their heterozygous siblings (I,J,K,L) were also double-labelled with sry-α antibody (I,K,M,O) and phalloidin (J,L,N,P). The sry-α protein in nullo mutants (M,O) does not form a hexagonal network at the level of the cellularization front. Bar in D (applies to entire figure), 10 µm. nullo protein in Drosophila embryos 1871 ison of the intracellular distributions of these two proteins is mental pathway. Previous experiments have shown that the merited. Mice were immunized with a truncated version of the genes are transcriptionally independent (Rose and Wieschaus, sry-α protein and supernatants from several of the resulting 1992). To determine if either protein regulates the translation monoclonal lines reacted with a protein from two- to three- or intracellular localization of the other, embryos derived from hour fly embryos of approximately 58 kDa, the size of sry-α either nullo or sry-α stocks were stained with antibodies to the as defined by polyclonal antisera (Schweisguth et al., 1990). reciprocal protein. The specificity of the monoclonal antibodies was shown by the By approximately 10 minutes into cycle 14, both nullo and failure of embryos deleted for the sry-α gene to exhibit any of sry-α mutant embryos can be distinguished from sibling the staining patterns described (see Fig. 7E,G). embryos: they exhibit an abnormal array of actin and myosin Although the sry-α protein has been reported to be present that has unusually thin and thick portions and some interfaces in cycle 13 embryos (Schweisguth et al., 1990), our antibodies that are completely disrupted. All cycle 14 embryos lacking only detect sry-α protein above background staining during the sry-α clearly contain nullo protein. The protein forms a formation of the hexagonal network in early cycle 14. As soon roughly hexagonal network in sry-α mutant embryos with as the hexagonal network of actin is present, sry-α forms a co- some disrupted interfaces (Fig. 6M,N,O,P). The discontinuities linear array (Fig. 5A,B). Like actin, the sry-α array is in nullo staining completely match the disruptions seen in the composed of ‘lines’ of sry-α staining with each line forming actin-myosin network. Therefore, nullo protein continues to be the side of a hexagon. However, the lines of sry-α protein often coincident with actin and myosin in the sry-α mutant back- do not meet at the hexagon vertices and sry-α is relatively ground. Some sry-α embryos appear to have fewer ‘dots’ of depleted there. This staining pattern is in marked contrast to nullo staining than wild-type embryos. It is not clear whether the dots of nullo staining that compose the sides and, most this decreased intensity of staining is due to an overall decrease prominently, the vertices of the hexagons. During early cellu- in levels of nullo protein or a slightly compromised ability of α larization stages (Fig. 5C,D), sry- appears to be associated the protein to be properly localized. Overall, however, the dis- α with the entire plasma membrane. The sry- protein shows an tribution of nullo protein is remarkably normal in sry-α additional distinctive localization that is not shared by actin, mutants. myosin or nullo: the protein is localized above each cycle 14 In contrast, absence of nullo activity has a more striking nucleus in a small spherical structure of unknown origin or sig- α α effect on the localization of sry- protein. While phenotypi- nificance (Fig. 5G,K). This sry- staining may be associated cally nullo embryos contain large pools of sry-α protein, its with the centrosomes or the microtubule arrays located apical distribution pattern is altered. sry-α protein fails to co-localize to the nuclei at these stages (Whitfield et al., 1988; Kellogg et with the leading edge of the cellularization front and and the al., 1989). α α well defined lines of sry- staining normally associated with Several basic features of sry- localization are maintained the actin-myosin array are absent (Fig. 7M,O). Instead, the throughout the slow phase (Fig. 5I,J,K,L,M,N,O,P) and much entire cytoplasmic region between the nuclei shows a low level of the fast phase (Fig. 5Q,R,S,T) of cellularization. The sry-α of sry-α staining. Localization of sry-α protein to the spherical protein remains associated with the invaginating actin-myosin structure above each nucleus persists, however, suggesting that hexagonal array until late in the fast phase. Although the actin- nullo activity is specifically required for association of sry-α myosin network assumes a ring-like appearance during fast protein with the actin-myosin network. phase, sry-α protein continues to form interlocking hexagons that are composed of lines of sry-α staining (compare Fig. 5M and 5N). In this regard, sry-α resembles nullo, since neither DISCUSSION protein co-localizes with the rings. It should be noted, however, that sry-α protein remains localized to the furrow Translation and post-translational modification of canal during more advanced stages of cellularization than nullo does. During cellularization, the association of sry-α with the the nullo protein plasma membrane is limited to the furrow canal and the lateral Transcription of nullo RNA was previously demonstrated to be membranes just apical to it. In addition, sry-α protein continues very tightly regulated (Rose and Wieschaus, 1992). The nullo to be localized to a spherical structure above each nucleus. No RNA is expressed only for a brief period during the Drosophila sry-α protein seems to be specifically localized to the apical life cycle: developmental Northern blots revealed the presence plasma membrane, but the protein is present in the cytoplasm of the nullo transcript in RNA samples from 0- to 4-hour above and, to an increasing extent, below the nuclei. Towards embryos but not in any other developmental stages. In RNA in the end of the fast phase of cellularization (Fig. 5S,T), the situs to whole mount embryos, the transcript was detectable localizations of sry-α protein to the cellularization front and to from the beginning of cycle 11 through the slow phase of cel- the spherical structures above the nuclei are lost and most of lularization. Within this short interval, large amounts of nullo the sry-α protein becomes dispersed throughout the cytoplasm. transcript rapidly accumulate, reaching a maximum level The intensity of sry-α antibody staining diminishes rapidly during the division between cycle 13 and 14. As soon as cycle during the final stages of cellularization, although sry-α protein 14 begins, levels of nullo RNA plummet. While accumulation remains detectable in gastrulating embryos (data not shown) of the nullo transcript occurs uniformly throughout the embryo, until the onset of germ band extension. nullo degradation does not; a reproducibly banded pattern of the nullo transcript is visible in early cycle 14. Localizations of nullo and sry-α in mutant embryos Both developmental western blots of individual, precisely The similarities in their mutant phenotypes suggest that nullo staged embryos and antibody staining of whole-mount and sry-α proteins may be components of the same develop- embryos indicate that the dynamics of nullo protein expression 1872 M. A. Postner and E. F. Wieschaus mirror those of its RNA. By western analysis, the nullo protein therefore provide a spatial cue, localizing other proteins into a is detectable from the start of cycle 13 through the beginning hexagonal array following the 13th mitosis. Since an actin- of gastrulation. The highest levels of nullo protein are detected myosin network, albeit abnormal, still forms in nullo mutants, in embryos in early cycle 14, just prior to and during the slow nullo protein cannot provide the only bias or scaffold for phase of cellularization. The levels of nullo protein then drop recruitment of microfilaments and myosin into the hexagonal rapidly during the fast phase of cellularization. The protein pattern. However, other components of the array might be expression pattern thus closely resembles the transcription completely reliant on cycle 13 nullo localization for their pattern with a translational lag of no greater than twenty proper distributions; lack of such proteins might cause the minutes. This comparison implies that nullo protein has a con- hexagonal array that forms in cycle 14 to be unstable and result sistently short half-life: nullo protein is constantly being in the disruptions seen in embryos lacking the nullo gene. degraded and replaced with newly synthesized protein until Given the similarities between the nullo and sry-α phenotypes, lack of transcript prevents its replacement. An additional an obvious possibility is the sry-α protein. mechanism of specific degradation late in cellularization In cellularizing embryos mutant for the nullo locus, the sry- cannot be ruled out. α protein is localized to a sphere above each nucleus and dis- Independent monoclonal antibodies specifically recognize tributed throughout the apical cytoplasm just as in wild-type two forms of the nullo protein in crude protein preparations embryos. However, the protein specifically fails to form a from early embryos. Preliminary results suggest that the size discrete hexagonal network. Instead, sry-α protein shows a difference between the two forms reflects a differential phos- cytoplasmic localization that is indistinguishable from that phorylation, since treatment with bacterial alkaline phos- seen in more apical regions. The failure of sry-α protein to phatase causes the conversion of the slower migrating form form even a disorganized hexagonal array in embryos lacking into the faster migrating form (Postner, 1993; and unpublished the nullo gene indicates that sry-α is dependent upon nullo for observations). The significance of this modification is unclear. its localization to the hexagonal network. On the other hand, We did not detect pronounced differences in the ratio of phos- nullo protein still co-localizes with the actin network in sry-α phorylated to unphosphorylated nullo protein during the entire mutants, even though that network is disrupted. This suggests window of nullo protein expression. that the relationship between the loci is not reciprocal and that α The nullo and sry-α proteins are components of the sry- protein is functionally downstream of nullo. The rela- α hexagonal network tively minor effect of sry- protein on final nullo distribution is probably indirect: a result of the instability of the actin- During early cycle 14, nullo and sry-α proteins both colocal- myosin network in sry-α mutants. ize to the hexagonal array of actin and myosin. They maintain In this model the nullo phenotype would result in part from this association at least until the fast phase of cellularization α (when the cleavage furrow has invaginated about half its final the failure of sry- to localize to the network and maintain its depth). A more thorough biochemical analysis is required to stability. This proposal does not account for the difference in determine whether either protein interacts directly with the severity between the two null phenotypes. Since more disrup- actin cytoskeketon. The distributions of the two proteins within tions of the hexagonal network occur in nullo mutant embryos α than in sry-α mutant embryos, the nullo phenotype cannot be the network are not identical. The sry- protein is found in α lines running along the sides of the hexagons. These lines due in its entirety to a mislocalization of sry- protein. In α rarely reach the hexagonal vertices and little sry-α protein is addition to facilitating the proper localization of the sry- detectable within the vertices. This contrasts with the distrib- protein, the nullo protein must further contribute to the stability ution of nullo protein, which shows a punctate pattern through- of the hexagonal array. It could do so indirectly by recruiting out the hexagonal array. nullo appears at regular intervals other stabilizing molecules to the network or directly by along the sides of the hexagons and is particularly abundant at binding, and thus anchoring, multiple components of the the hexagonal vertices. The punctate staining pattern is network. observed from the earliest stages when the protein can be detected and suggests a vesicular localization at least during We thank Dan Kiehart for the gifts of antibodies and Marty Marlow of the Princeton’s monoclonal facility for establishing the antibody synthesis or transport to the surface. The potential myristoyla- lines. Joe Goodhouse, Mark Peifer and Romy Knittel provided tion codon at the amino terminus of the nullo protein (Rose valuable technical advice on confocal microscopy, fusion proteins and and Wieschaus, 1992) might provide a mechanism for associ- monoclonal antibody screening, respectively. We are indebted to ating the protein with transport vesicles and ultimately with the Alain Vincent for supplying the construct from which the truncated plasma membrane. The discovery that the Drosophila virilis sry-α protein was produced and for sharing unpublished data and homologue of the nullo protein contains an equally favorable polyclonal antisera to sry-α with us. This work was supported by grant potential myristoylation codon amidst an otherwise divergent 5RO1 HD15587 from the National Institutes of Health. amino terminus (E. Schejter, personal communication) strongly suggests that this sequence is important functionally. If the nullo protein is myristoylated in vivo, nullo might also REFERENCES provide a crucial link between the actin hexagonal network and the plasma membrane. Arking, R. and Parente, A. (1980). Effects of RNA inhibitors on the During interphase of cycle 13, nullo forms a network of development of Drosophila embryos permeabilized by a new technique. J. Exp. Zool. 212, 183-194. interconnecting rings of protein that is associated with the Edgar, B. A. and Schubiger, G. (1986). Parameters controlling transcriptional surface regions between the actin caps. This hexagonal pattern activation during early Drosophila development. Cell 44, 871-877. precedes formation of the actin hexagonal array and might Foe, V. E. and Alberts, B. M. (1983). Studies of nuclear and cytoplasmic nullo protein in Drosophila embryos 1873

behavior during the five mitotic cycles that precede gastrulation in Simpson, L. and Wieschaus, E. (1990). Zygotic activity of the nullo locus is Drosophila embryogenesis. J. Cell Sci. 61, 31-70. required to stabilize the actin-myosin network during cellularization in Harlow, E. and Lane, D. (1988). Antibodies, A Laboratory Manual. Cold Drosophila embryos. Development 110, 851-863. Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Smith, D. B. and Johnson, K. S. (1988). Single-step purification of Ibnsouda, S., Schweisguth, F. de Billy, G. and Vincent, A. (1993). polypeptides expressed in Escherichia coli as fusions with glutathione S- Relationship between expression of serendipity-α and cellularization of the tranferase. Gene 67, 31-40. Drosophila embryo as revealed by intraspecific transformation. Studier, F. W. and Moffat, B. A. (1986). Use of bacteriophage T7 RNA Development 119, 471-483. polymerase to direct selective high-level expression of cloned genes. J. Mol. James, A. A. and Vincent, A. (1986). The spatial distribution of a blastoderm Biol. 189, 113. stage-specific mRNA from the serendipity locus of Drosophila Turner, F. R. and Mahowald, A. P. (1976). Scanning electron microscopy of melanogaster. Dev. Biol. 118, 474-479. Drosophila embryogenesis. I. The structure of the egg envelopes and the Kellogg, D. R., Field, C. M. and Alberts, B. M. (1989). Identification of formation of the cellular blastoderm. Dev. Biol. 50, 95-108. microtubule-associated proteins in the centrosome, spindle, and kinetochore Vincent, A., Colot, H. V. and Rosbash, M. (1985). Sequence and structure of of the early Drosophila embryo. J. Cell Biol. 109, 2977-2991. the serendipity locus of Drosophila melanogaster. A densely transcribed Kiehart, D. P., Ketchum, A., Young, P., Lutz, D. and Alfentino, M. R. S. region including a blastoderm-specific gene. J. Mol. Biol. 186, 149-166. (1990). Contractile proteins in Drosophila development. Ann. NY Acad. Sci. Warn, R. M., Bullard, B. and Magrath, R. (1980). Changes in the distribution 582, 233-251. of cortical myosin during the cellularization of the Drosophila embryo. J. Mabuchi, I. (1986). Biochemical aspects of cytokinesis. Int. Rev. Cytol. 101, Embryol. Exp. Morph. 57, 167-176. 175-213. Warn, R. M. and Magrath, R. (1983). F-actin distribution during the Merrill, P. T., Sweeton, D. and Wieschaus, E. (1988). Requirements for cellularization of the Drosophila embryo visualized with FL-phalloidin. Exp. autosomal gene activity during precellular stages of Drosophila Cell Res. 143, 103-114. melanogaster. Development 104, 495-509. Warn, R. M. and Robert-Nicoud, M. (1990). F-actin organization during the Postner, M. (1993). Developmental genetics of cytoskeletal and cytoplasmic cellularization of the Drosophila embryo as revealed with a confocal laser reorganizations in the Drosophila melanogaster blastoderm embryos. Ph.D. scanning microscope. J. Cell Sci. 96, 35-42. thesis, Princeton University. Whitfield, W. G. F., Miller, S. E., Saumweber, H., Frasch, M. and Glover, Rose, L. S. and Wieschaus, E. (1992). The Drosophila cellularization gene D. M. (1988). Cloning of a gene encoding an antigen associated with the nullo produces a blastoderm-specific transcript whose levels respond to the centrosome in Drosophila. J. Cell Sci. 89, 467-480. nucleocytoplasmic ratio. Genes Dev. 6, 1255-1268. Wieschaus, E. and Sweeton, D. (1988). Requirements for X-linked zygotic Salmon, E. D. (1989). Cytokinesis in animal cells. Curr. Opin. Cell Biol. 1, activity during cellularization of early Drosophila embryos. Development 541-7. 104, 483-493. Satterwhite, L. L. and Pollard, T. D. (1992). Cytokinesis. Curr. Opin. Cell Young, P. E., Pesacreta, T. C. and Kiehart, D. P. (1991). Dynamic changes in Biol. 4, 43-52. the distribution of cytoplasmic myosin during Drosophila embryogenesis. Schejter, E.D. and Wieschaus, E. (1993). bottleneck acts as a regulator of the Development 111, 1-14. microfilament network governing cellularization of the Drosophila embryo. Zalokar, M. and Erk, I. (1976). Division and migration of nuclei during early Cell 75, 373-385. embryogenesis of Drosophila melanogaster. J. Microsc. Biol. Cell 25, 97- Schroeder, T. E. (1990). The contractile ring and furrowing in dividing cells. 106. Ann. NY Acad. Sci 582, 78-87. Schweisguth, F., Lepesant, J.-A. and Vincent, A. (1990). The serendipity- alpha gene encodes a membrane-associated protein required for the cellularization of the Drosophila embryo. Genes Dev. 4, 922-931. (Received 4 February 1994 - Accepted 11 April 1994)