BIOLOGICAL OVERVIEW

Cellularization of the Drosophila embryo is a specialized form of cytokinesis that couples membrane growth with the formation of a polarized epithelium. slow as molasses (slam) is essential for polarized growth of the plasma membrane during cellularization. In slam mutant embryos, the furrow canal is disorganized, and polarized insertion of transmembrane proteins is disrupted. slam shows a striking developmental induction during the slow phase of cellularization, and Slam protein localizes to the furrow canal and the basal junction. Slam colocalizes with the junctional proteins Armadillo/ß-catenin, the PDZ domain-containing protein Dlt (renamed Patj), and Myosin (see Zipper) and is also required for their proper membrane localization. These results suggest that developmental induction of Slam organizes the polarized growth of membrane via the recruitment of membrane-targeting proteins at adherens junctions (Lecuit, 2002 and Stein, 2002).

slam is also required for germ cell migration. In slam zygotic mutants, germ cells fail to transit off the midgut into the mesoderm. slam is required at this stage in parallel to HMG Coenzyme A reductase, another germ cell migration gene. Because slam RNA and protein are expressed earlier than the time when defects are observed in germ cell migration, it is proposed that Slam is required for the localization of a signal to the basal side of blastoderm cells that is needed later in the posterior midgut to guide germ cells (Stein, 2002).

Cleavage of the Drosophila syncytial embryo is known as cellularization and provides a striking example in which morphogenetic processes are associated with profound rearrangements of the plasma membrane. During cellularization, the plasma membrane surface grows about 30-fold, invaginates between cortical nuclei, and produces a polarized epithelium. The first 13 nuclear divisions occur in a single cytoplasm and result in about 5000 nuclei located at the cortex of the embryo. At the beginning of cycle 14, the divisions cease, and the membrane invaginates between each nucleus, partitioning them into single unit cells. The first 35-40 min of cellularization (slow phase) account for the first 10 μmmeters of new membrane along what will become the basal-lateral surface of each cell. The speed of invagination rapidly increases during the last 15-20 min (fast phase) and stops when the invaginating front is located about 30 μmeters inside the embryo. By using pulse labeling of the plasma membrane, it has been shown that membrane growth during cellularization involves the regulated mobilization of membrane pools from the secretory pathway to precisely defined sites (Lecuit, 2000). The modes of membrane growth are different in slow phase and fast phase, paralleling the change in the rate of membrane invagination. During slow phase, membrane growth is concentrated apically. When the plasma membrane is labeled with a fluorescent lectin before membrane invagination begins, the apical label is rapidly lost, whereas it persists and even accumulates in the region that will form the front of invagination called furrow canal (FC). The same pattern of membrane dynamics is maintained as the embryo progresses through the end of slow phase; apical membrane is rapidly replaced with intracellular unlabeled membrane, whereas the basal-lateral membrane remains labeled as it forms and grows. This observation suggests that the apical membrane is the principle site of new unlabeled membrane insertion and that, during cellularization, a lateral domain with different membrane turnover properties is established adjacent to the original apical surface. This lateral domain is apparent prior to invagination and persists during slow phase. During fast phase, a new site of membrane insertion is added in an apical-lateral region of the plasma membrane (Lecuit, 2002).

In order to understand the mechanisms underlying the regulated membrane addition during cellularization, a search was carried out for genes required for stage-specific membrane growth. Drosophila cellularization occurs at a key developmental transition similar to the mid blastula transition in Xenopus (MBT) and is characterized by the induction of zygotic gene expression. Two categories of gene products contribute to membrane invagination during cellularization: maternally supplied RNAs and proteins deposited by the mother during oogenesis and zygotic gene products induced during cellularization. Genetic screens have identified a surprisingly low number of genomic regions required zygotically for cellularization, suggesting that less than ten nonredundant genes might control the entire process (Merrill, 1988; Wieschaus, 1988). This finding suggests that such zygotic genes might correspond to key regulators whose expression at specific points during cellularization activates a maternal machine poised to deliver membrane pools to specific sites of the plasma membrane (Lecuit, 2002).

A new gene, slow as molasses, has been identified that is required for the growth of the basal-lateral membrane during cellularization. slam expression is rapidly induced during slow phase at the beginning of cycle 14. In slam mutant embryos, the membrane that would normally form the furrow canal and basal lateral membrane never assumes the smooth morphology devoid of villous projections characteristic of that region. The furrow canal and basal junction do not form. This morphological defect is accompanied by defects in apical membrane trafficking, such that transmembrane proteins like Neurotactin (Nrt) and Toll accumulate abnormally in the apical cytoplasm. slam encodes a novel protein that localizes in a polarized fashion to the furrow canal and to the basal adherens junction but not apically. Slam colocalizes with and is required for the proper membrane accumulation of junctional components such as ß-catenin/Armadillo (Arm) and the PDZ domain protein Patj and of Myosin. Continued expression of slam after cellularization leads to a polarized distribution of the protein in the apical adherens junctions along the anterior and posterior borders of cells. In such embryos, Slam recruits Myosin to the junction in a similar asymmetric pattern, arguing that the two proteins interact in vivo. It is proposed that Slam organizes the polarized growth of the basal-lateral membrane via the regulation of membrane insertion at the level of basal junctions. These findings substantiate the notion that morphogenesis and polarization of the plasma membrane are inherently linked processes (Lecuit, 2002).

Their small number and the specificity of their phenotypes suggest that genes required zygotically for cellularization may define key steps regulating or inducing specific aspects of the process. Previous experiments have identified different patterns of membrane insertion during the slow and fast phases of cellularization. During slow phase, growth of the membrane is concentrated apically. But, during fast phase, a new site of membrane insertion is superimposed apicolaterally. This striking observation suggests that the subdivision of the entire process based on the kinetics of invagination might also reveal distinct mechanisms of membrane insertion and growth. If so, one predicts that genes specifically regulating membrane growth in one phase or the other ought to exist. Several features of slam are consistent with such a specific role. In slam mutant embryos, membrane growth is inhibited during slow phase, while fast phase appears at the right timing. In addition, the induction of slam RNA correlates with that of slow phase. Slam levels peak at the beginning of cycle 14, when slow phase begins. The RNA is then degraded ~40 min later during fast phase. Finally, slam controls the formation of membrane structures that are specific to cellularization: the furrow canal and its associated basal junction as well as the basal-lateral membrane surface. It is likely that Slam expression induces the transition to slow phase and thus controls the onset of cellularization (Lecuit, 2002).

Because slam RNA (and potentially Slam protein) persist when fast phase is initiated, the transition to fast phase may involve the addition or superimposition of a second membrane insertion pathway, such that the observed increase in membrane invagination may reflect the sum of two pathways. This would explain the reduced rate of membrane invagination during fast phase in slam(RNAi) embryos.

slam appears to be unique in the genome in respect to its requirement during slow phase. RNA(i) to slam reproduces the phenotype of embryos deleted for almost the entire left arm of the second chromosome. Translocation screens have identified no other regions that affect membrane growth during the slow phase of cycle 14 (Lecuit, 2002).

slam, however, is not the only gene required zygotically at the beginning of cycle 14: other regions of the genome affect cellularization, although the associated phenotypes are morphologically distinct from those of slam. For example, expression of the nullo gene is required for the stabilization of a basal adhesive junction that isolates the furrow canal and allows the stable accumulation of Myosin (Hunter, 2000). Because slam mutants also show decreased Myosin levels in the furrow canal and fail to form basal junctions, it is possible that both genes work in concert to establish polarized membrane insertion and extension at the level of junctions. Slam may determine the specific character of the protein targeting during slow phase, but the consequences may depend on the existence of specific junctions and other aspects of membrane structure. It is therefore difficult to predict the phenotype of slam expression in stages that do not have the same constellation of factors present at cycle 14. It has not been possible to express Slam earlier than cycle 14 and thus it is not known whether such expression would be sufficient to induce premature cellularization (Lecuit, 2002).

Slow phase is characterized by the formation of two adjacent membrane domains, as revealed by data using pulse labeling of plasma membrane and structural data using electron microscopy. Together, one region of the membrane containing many villous projections has been identified; here lectin label is rapidly removed. In a second adjacent smooth membrane domain, lectin persists and accumulates during slow phase. This early polarity of the plasma membrane is also revealed by the distinct localization of various proteins in the smooth area of the membrane that becomes the furrow canal during slow phase: Dlt (now Patj) and Slam. Note that the same basic pattern persists as invagination, per se, is initiated. The growing lateral membrane is smooth and keeps a high level of labeled membrane in contrast to the apical membrane. Two mechanisms can be envisioned to account for this partitioning of two membrane domains as the basal-lateral membrane grows. (1) As proposed previously (Lecuit, 2000), the newly inserted membrane shows little miscibility with the recipient apical membrane. Such limited mixing has been reported in other systems. Slam's localization in the well organized smooth area of membrane and its exclusion from the area rich in villous projections is consistent with this model. (2) Slam could support a membrane-based scaffold that stabilizes that region and keeps its integrity as new membrane is inserted. The observation that Slam colocalizes with Myosin and the PDZ domain-containing protein Dlt/Patj and that it is required for their proper membrane accumulation lends further support to this view. The induction of Slam could induce the formation of a protein complex that connects the actin-Myosin cytoskeleton to the plasma membrane via Dlt/Patj. PDZ domain-containing proteins such as Dlg/Patj have indeed been implicated in the clustering of transmembrane proteins, such as Fasciclin, and the ion channel Shaker in the synapses (Lecuit, 2002 and references therein).

slam-dependent formation of the furrow canal and of the basal-lateral membrane as a separate membrane domain could alternatively involve an active intracellular membrane transport mechanism. The rapid removal of apically labeled membrane and its accumulation basal laterally could indeed be tightly linked if an endocytic route transfers apical membrane laterally in a manner akin to transcytosis. This transfer of membrane would involve traffic through and sorting from endosomes and intersect the exocytic pathway. Transcytosis is a well-known pathway required for the polarization of hepatocytes, for instance. Slam could stabilize or enhance the junctional insertion of membrane basal laterally and therefore favor the polarized assembly of the lateral membrane. The site of membrane integration and assembly could be the basal junction area, since junctions are known to recruit proteins, such as sec6/sec8, required for the basal-lateral targeting of transport vesicles in MDCK cells and for the associated growth of the basal-lateral membrane. Consistent with this proposal is the finding that the subcellular localization of integral membrane proteins such as Toll and Nrt is abnormal in slam mutant embryos. While these proteins are clearly in part inserted in the apical plasma membrane, they also accumulate in a diffuse apical pattern very distinct from their normal well-defined plasma membrane accumulation. This diffuse cytoplasmic accumulation could not be resolved into punctate structures or well-defined organelles. One possibility is that, in the absence of Slam, the lateral transfer of membrane containing Toll and Nrt is highly inefficient and not polarized at all. Instead of accumulating laterally and contributing to the growth of a distinct membrane domain, vesicles may be routed back to the apical membrane but partially accumulate in the apical cytoplasm. In such a scenario, Slam's unique developmental induction stabilizes the polarized assembly of membrane basal laterally. The suggested interaction between Slam, Myosin, and Dlt/Patj at the level of junctions might be involved in the assembly of membrane-targeting complexes that connect the plasma membrane with the actin cytoskeleton. Biochemical experiments that will identify molecular partners of Slam should be particularly revealing. Although it is not possible to distinguish between these two possibilities, the latter is favored because, unlike the former, it accounts for the observed defects in the localization of Nrt and Toll (Lecuit, 2002).

In conclusion, slam encodes a developmental regulator of membrane morphogenesis during cleavage of the Drosophila embryo. Slam affects the polarized growth of the basal-lateral membrane and its organization as a distinct membrane domain. The data therefore substantiate the notion that membrane growth and polarity are indeed coregulated processes in epithelial cells and that junctions play an important role in this process. Future experiments will reveal whether this also involves a regulation of polarized targeting or not. Slam may prove a very useful entry point into the mechanisms of epithelial polarization and morphogenesis during development (Lecuit, 2002).

Function and dynamics of slam in furrow formation in early Drosophila embryo

The Drosophila embryo undergoes a developmental transition in the blastoderm stage switching from syncytial to cellular development. The cleavage furrow, which encloses nuclei into cells, is a prominent morphological feature of this transition. It is not clear how the pattern of the furrow array is defined and how zygotic genes trigger the formation and invagination of interphase furrows. A key to these questions is provided by the gene slow as molasses (slam), which has been previously implicated in controlling furrow invagination. This study investigated the null phenotype of slam, the dynamics of Slam protein, and its control by the recycling endosome. slam was found to be essential for furrow invagination during cellularisation and together with nullo, for specification of the furrow. During cellularisation, Slam marks first the furrow, which is derived from the metaphase furrow of the previous mitosis. Slightly later, Slam accumulates at new furrows between daughter cells early in interphase. Slam is stably associated with the furrow canal except for the onset of cellularisation as revealed by FRAP experiments. Restriction of Slam to the furrow canal and Slam mobility during cellularisation is controlled by the recycling endosome and centrosomes. A three step model is proposed. The retracting metaphase furrow leaves an initial mark. This mark and the border between corresponding daughter nuclei are refined by vesicular transport away from pericentrosomal recycling endosome towards the margins of the somatic buds. Following the onset of zygotic gene expression, Slam and Nullo together stabilise this mark and Slam triggers invagination of the cleavage furrow (Acharya, 2013).

Drosophila cellularisation is a specialised form of cytokinesis and transforms the syncytial into a cellular blastoderm. Embryonic development starts with a series of 13 rapid nuclear divisions that take place in a common cytoplasm without cytokinesis. In interphases of the syncytial divisions 10-13, the nuclei together with their associated centrosomes shape the surface of the overlaying embryonic plasma membrane by triggering the formation of actin-rich somatic buds. During mitosis, transient invaginations of the plasma membrane, so-called metaphase furrows, extend towards the interior of the embryo to separate adjacent mitotic spindles. These metaphase furrows retract during telophase. A developmental switch occurs after the last nuclear division at the onset of interphase 14. The plasma membrane starts to invaginate at the margins of adjacent somatic buds. At the site of prospective invagination, the membrane transforms into a hairpin-like canal, which is called furrow canal. It is unclear whether the interphase furrow forms de novo or builds on information derived from the metaphase furrow. Concomitantly to furrow formation, the plasma membrane becomes polarised (Acharya, 2013).

Genetic analysis showed that two processes, organisation of the cytoskeleton and membrane trafficking, largely ensure proper invagination of the plasma membrane during cellularisation. Cytoskeleton organisation is controlled by factors such as Rho1, RhoGEF2, Dia and Abl. Membrane trafficking includes the polarized insertion of new plasma membrane to permit the enormous increase in membrane surface as well as the regulation of endo- and exocytosis. Not surprisingly, many key regulators of membrane trafficking have an important function during cellularisation. For example, the recycling endosome, which is controlled by rab11 and nuf, is required for cellularisation and accumulation of RhoGEF2 (Acharya, 2013).

As these components are present throughout early development, they do not trigger furrow invagination in interphase 14. Zygotic genes are likely candidates for a trigger. Genes such as nullo, slam, bottleneck and sry-α are candidate genes. Previous studies revealed a role of slam in furrow invagination (Lecuit, 2002; Stein, 2002) and recruitment of Patj and RhoGEF2 (Wenzl, 2010). Mutations in nullo and sry-α are characterised by an incomplete hexagonal membrane array and the presence of multinuclear cells. Furthermore, nullo controls the separation of lateral and basal compartments and actin-dependent stabilisation of the basal membrane (Acharya, 2013).

Whereas the mechanisms underlying the invagination process have been intensively studied over the past few years, the initial events of furrow invagination are still poorly understood. Slam is key to understanding furrow invagination. Slam is an early marker of the furrow and is required for proper furrow invagination. As only an initial analysis of the slam function has been previously reported, this study defined the null phenotype of slam and investigated the dynamics and spatial restriction of Slam protein (Acharya, 2013).

A first surprising finding of this study was the essential role of slam for furrow invagination but not furrow specification. Based on genetic analysis of available alleles and RNAi injection, it has been previously proposed that slam promoted the speed of furrow invagination. By generating a slam deletion, this study demonstrated that slam not only promotes but is essential for furrow invagination. Initial furrow formation and hexagonal arrangement is specified in the absence of slam despite the lack of a morphologically visible furrow. A possible explanation for this finding is that slam is not involved in defining this site, although Slam constitutes an early marker. Alternatively, slam may act together with another zygotic gene, namely nullo. This study presents data that are consistent with the second model. nullo slam double mutants lose the organisation of the furrow array as revealed by staining for Dia and E-CadherinGFP. These experiments demonstrate that nullo and slam have redundant functions. It is difficult to judge whether the disorganised Dia staining in the double mutants reflects a disorganised furrow. If this were the case, a third zygotic input beside Slam and Nullo would be necessary. Alternatively, the disorganised Dia staining may reflect cortical localisation of Dia and tendency to aggregate. Slam and Nullo are the main signals for specification of the furrow array in this scenario. In any case, it is clear that slam and nullo collaborate for an early aspect of cellularisation in addition to their distinct functions. Such an early function of nullo is surprising, as nullo embryos show very mild defects at low temperature. The strong phenotype of the double mutant is reminiscent to the loss of hexagonal pattern in embryos with depleted F-actin by injection of latrunculin. These findings are also consistent with previous reports that implicated nullo and downstream targets of slam in F-actin regulation (Acharya, 2013).

A second surprising and unexpected finding was a switch in Slam dynamics from a high FRAP recovery rate in mitosis/early cellularisation to slow and incomplete recovery throughout cellularisation. Other proteins at the furrow canal such as the PDZ-domain of RhoGEF2 and Amphiphysin-YFP completely exchange within a minute during cellularisation. The recovery of GFPslam fluorescence may be due to three distinct mechanisms: (1) exchange of bleached molecules at the furrow canal with unbleached molecules from the surrounding, (2) active transport of vesicles or particles loaded with Slam, (3) localised translation. It is necessary to consider localised translation as a potential mechanism. slam RNA and protein co-localise at the furrow canal. This colocalisation allows for localised protein translation restricted to the site of protein localisation (Wenzl, 2010). As no flow of Slam punctae was detected, neither when the 'new' furrow formed nor in the FRAP experiment, the second model for fluorescence recovery is not favored. Presently, there is no data that would allow distinguishing the models 1 and 3. Future experiments with photo-convertible tags and inhibition of translation by cycloheximide will allow the potential role of localised translation to be addressed (Acharya, 2013).

The molecular determinants for the switch in recovery behaviour of Slam are unknown, as there are no observed indications for posttranslational modifications. However, as the exchange rate is increased in nuf mutants, the recycling endosome may be involved, possibly in an indirect manner. It is conceivable that restriction of Slam to the basal domain leads to a stably bound population of Slam. In contrast, unrestricted plasma membrane localisation of Slam may be based on a less stable association. It is worth noting that Slam seems to have an intrinsic affinity to membranes as Slam expressed in cultured S2 cells is cortically enriched (Wenzl, 2010). Slam restriction to the basal compartment may enhance and stabilise this membrane affinity (Acharya, 2013).

Thirdly, it was unexpected to find differential labelling of 'old' and 'new' furrows by GFPslam. Such a labelling dynamics has not been described previously. For example, F-actin marks the metaphase furrow and cellularisation furrow but does not allow distinguishing 'old' and 'new' furrows. Similarly, E-CadherinGFP labelling is not different in 'old' and 'new' furrows. The continuous GFPslam labelling of the 'old' furrow and delayed and shi dependent accumulation at the 'new' furrow suggests that there are two mechanisms for initial definition of the furrow pattern: (1) a mechanism that uses the existing information from the previous cycle, (2) a mechanism that fills the gaps in the hexagonal furrow array between the respective daughter nuclei, similar to conventional cytokinesis. The different properties of 'old' and 'new' furrows become obvious by their differential dependence on vesicular trafficking, as revealed in shi mutants. de novo accumulation of Slam at 'new' furrows depends on vesicular budding, whereas 'old' furrows are not affected under the experimental conditions. The situation was not investigated during syncytial interphases, as slam has no function in these cycles. The maternally expressed GFPslam, however, marks a pseudo-hexagonal pattern during these cycles. This syncytial staining pattern suggests that the pattern-forming process is maternally determined and that a potential Slam receptor is already present in syncytial embryos (Acharya, 2013).

These data support and further define a previously proposed model. It has been proposed that vesicle transport of the recycling endosome would be organised by the centrosomes. A pseudo-hexagonal pattern would emerge from the regular distribution of centrosomes. Endosomal uptake and targeted exocytosis would lead to restriction of a membranous Slam receptor to a sharply defined domain within the plasma membrane at maximal distance from the respective centrosomes. The hexagonal pattern would be completed by de novo formation of a 'new' furrow situated between corresponding daughter nuclei. After the onset of zygotic gene expression in interphase 14, the marks would be used to accumulate new zygotic proteins such as Slam and Nullo, which would define the basal domain and trigger membrane invagination and F-actin accumulation. In addition to these functions, Slam and Nullo would maintain the furrow structure and define the basal domain. The model predicts that premature expression of the essential zygotic genes triggers membrane invagination already in syncytial interphases. No signs of membrane invagination were observed in embryos maternally expressing GFPslam. However, precocious onset of many, if not all zygotic genes can induce cellularisation already in interphase 13. It will be interesting to see which set of zygotic genes suffices to trigger furrow formation in syncytial cycles (Acharya, 2013).

GENE STRUCTURE

cDNA clone length - 4093

Bases in 5' UTR - 379

Exons - 3

Bases in 3' UTR - 123

PROTEIN STRUCTURE

Amino Acids - 1196

Structural Domains

A complete slam cDNA identified by the Genome Project comprises an open reading frame 3591 nucleotides long and encoding a large putative 1196 amino acid protein (135 kDa) without any homolog in the database. No signal sequence or putative transmembrane domain is found, arguing that the Slam protein might be cytosolic. BLAST searches fail to identify any conserved protein domain, except for a potential coiled-coil motif between amino acids 516 and 546 and low similarity to protein phosphatases (Lecuit, 2002).

date revised: 5 June 2004

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.