End4/Sla2 is involved in establishment of a new growth zone in Schizosaccharomyces pombe

Many cellular activities including motility, differentiation and division, depend on the ability of cells to become polarized, a process that generally requires assembly of a polarized actin cytoskeleton to target secretion, vesicular transport and growth. Understanding how this polarization is brought about is a key problem in cell biology.

In the rod-shaped yeast Schizosaccharomyces pombe, polarity is expressed in the pattern of cell growth, which is restricted to the ends of the long axis of the cell. This controlled spatial organization depends on both the actin and microtubule cytoskeletons. F-actin is required for polarized growth, whilst the microtubular cytoskeleton is required to accurately position the growth zones at the opposite ends of the cell. Actin is organized in cortical patches associated with the growing surfaces of the cell and interference with the actin cytoskeleton completely disrupts cell polarity, resulting in cells becoming round. In contrast, disruption of the microtubule cytoskeleton results in mispositioning of the growth zones and leads to alterations in cell morphology (Hagan, 1998; Sawin and Nurse, 1998). During interphase, microtubules are organized in antiparallel bundles with the plus ends facing the tips of the cell where they deliver the polarity marker, Tea1 (Drummond and Cross, 2000; Hagan, 1998; Mata and Nurse, 1997). Cells lacking Tea1 still polarize growth but either fail to properly position the growth zones, which results in bending and branching of cells, or grow only from one end, leading to monopolar growth (Mata and Nurse, 1997).

In wild-type cells the pattern of growth changes during the cell cycle. After cell division growth is monopolar, taking place exclusively at the `old' end, which was present in the previous cell cycle. However once a minimal cell size requisite is met and DNA replication is completed, growth is activated at the opposite `new' end that was generated at cytokinesis. This transition from monopolar to bipolar growth is known as new end take off or NETO (Mitchison and Nurse, 1985), and represents a major morphogenetic transition in the fission yeast cell cycle. We reasoned that understanding the mechanisms underlying NETO might reveal new links between the actin and microtubule cytoskeletons and illuminate their roles in the spatial organization of the cell. Hence, we looked for homologues of known regulators of actin dynamics that might link actin organization to new end localization.

One such regulator is Sc-SLA2, which is essential in budding yeast for the proper assembly and function of the cortical actin cytoskeleton and for the establishment of cell polarity (Holtzman et al., 1993). Sc-SLA2 binds actin in vitro through its talin-like domain and partially colocalizes with cortical actin patches at the growing surface of the cell (McCann and Craig, 1997; Yang et al., 1999). Loss of Sc-SLA2 activity results in complete delocalization of cortical actin patches (Holtzman et al., 1993) as well as defects in the internalization step of endocytosis (Kaksonen et al., 2003; Raths et al., 1993). Homologues of Sc-SLA2 have already been identified in humans as huntingtin interacting proteins (Hip1 and Hip1R), and it has been suggested that Sla2-related proteins link the endocytic machinery to the actin cytoskeleton (Engqvist-Goldstein et al., 1999). By interfering with membrane-actin cytoskeleton interactions (Kalchman et al., 1997), the loss of the huntingtin-Hip1R interaction might contribute to the neurodegenerative disorder observed in Huntington patients.

Here we show that the recently identified S. pombe orthologue of Sc-SLA2 (End4/Sla2) (Iwaki et al., 2004) is enriched at sites of actin accumulation in fission yeast, is required for polarized cell growth and organization of the cortical actin cytoskeleton and its localization at the new end depends on Tea1. We conclude that End4/Sla2 is required at the new end to organize or activate a new growth zone.

All S. pombe strains used in this study are listed in Table 1. Standard methods were used for growth, transformation and genetic manipulations (Moreno et al., 1991). All experiments, unless otherwise stated, were performed in YES (yeast extract with added 250 mg/l histidine, adenine, leucine and uridine). sla2Δ, sla2-GFP and sla2Δtalin were generated using a PCR-based method (Bahler et al., 1998) with a kanMX template and oligos with 60-80 nt long homology to sla2. Sla2R1083G was generated using the same approach, with an oligonucleotide bearing the point mutation. All deletions, mutations and tagging were performed in a diploid strain and then sporulated to produce viable haploids. Deletions and insertions were checked by PCR for presence of a KanMX cassette at the right locus. Point mutants were confirmed by sequencing of genomic DNAs.

For actin staining, cells were fixed by adding formaldehyde (with a final concentration of 3.7%) to the medium for 25 minutes, washed twice with PEM (100 mM Pipes, 1 mM EGTA, 1 mM MgSO₄, pH 6.9), permeabilized with 1% Triton X-100, washed twice with PEM and stained with Alexa Fluor 488-phalloidin (Molecular Probes).

FM4-64 staining was performed as described by Brazer et al. (Brazer et al., 2000). Briefly, cells were pelleted and resuspended in YES containing 10 nM FM4-64 (Molecular Probes). Cells were incubated at 4°C for 30 minutes, then washed, resuspended in fresh YES medium at a concentration of 5×10⁶ cells/ml and grown at 25°C. Staining was monitored every 5 minutes for up to 2 hours using an Axioplan Zeiss microscope (546 nm).

For immunofluorescence, cells were collected by filtration and fixed in –80°C methanol for 8 minutes and processed as described previously (Alfa et al., 1993). For microtubule detection, TAT1 (α-tubulin; a kind gift from K. Gull) antibody was used at 1:200 dilution, and Alexa Fluor 546-linked anti-mouse (Molecular Probes) at 1:1000 was used as secondary antibody. Photographs were taken using an Axioplan Zeiss microscope.

Cells were grown at 25°C in YES to OD₅₉₅ 0.5. Methyl-2-benzimidazole-carbamate (MBC) and brefeldin A (BfA) were then added to the medium to a final concentration of 25 μg/ml and 100 μg/ml, respectively. Sla2-GFP-expressing cells were fixed with –80°C methanol and stained for tubulin. Fluorescence from Sla2-GFP is very strong and was still visible after fixation. For latranculin A (latA) treatment, Sla2-GFP-expressing cells were fixed to a coverslip coated with lectin (100 μg/ml). Medium containing 40 μM latA was flushed under the coverslip and time-lapse images were taken every 2 seconds.

For block and release experiments, cdc10-129 cells were blocked for 3 hours (unless otherwise stated) at the restrictive temperature (36°C). Following drug addition (at appropriate concentrations as previously stated), cells were quickly released to the permissive temperature (25°C) and samples for immunofluorescence and Sla2-GFP were taken at different time points. For latA treatment, cells were filtered following the block and resuspended in 1/10 of the original volume in medium containing 40 μM latA. After 10 minutes cells were released to permissive temperature and stained after 30 minutes.

LatA pulse experiments were performed as described by Rupes et al. (Rupes et al., 1999). Briefly, cells were blocked at restrictive temperature for 3 hours. Aliquots were filtered and resuspended in 1/10 volume of medium containing 40 μM latA for 6 minutes at 36°C. LatA was then washed out by filtration, cells were washed with preconditioned medium and resuspended in fresh medium. Cells were fixed for actin staining at 30-minute time intervals.

We identified the sla2 gene in the S. pombe genome database by homology with the S. cerevisiae sla2 gene. S. pombe sla2 (SPAC688.11) encodes a protein with a predicted molecular mass of 123 kDa. Sequence alignment with other Sla2-related proteins showed that End4/Sla2 shares 36% similarity with Sc-SLA2 and 26% similarity with human HIP1R. End4/Sla2 has the same domain structure and organization as the other Sla2-related proteins, namely an N-terminal ENTH (epsin N-terminal homology) domain, a central coiled coil region, and a C-terminal talin-like domain. The talin-like domain contains the I/LWEQ-module predicted to bind to F-actin (McCann and Craig, 1999).

sla2 was deleted in a wild-type diploid using a PCR-based approach (Bahler et al., 1998). As in S. cerevisiae, sla2Δ cells were temperature sensitive for growth; cells failed to grow at 36°C and at 32°C, and grew slowly at 25°C with a doubling time of about 6 hours, more than twice that of wild-type cells (2 hours 50 minutes). sla2Δ cells grown at 25°C were wider than wild-type cells and showed a variety of growth patterns: cells were often branched, swollen in their middle regions or exhibited a complete loss of polarity (Fig. 1).

sla2Δ cells had an aberrant actin cytoskeleton. In wild-type interphase cells, F-actin is organized in polarized cortical patches enriched at growing tips and in thin cables running along the main axis of the cell (Marks and Hyams, 1985). In contrast, about half of sla2Δ cells at 25°C showed misoriented cables (data not shown) and disorganized cortical structures, with actin patches delocalized throughout the cell body (Fig. 1C). In the cells, which maintained the normal rod shape at 25°C, actin patches were observed only at one end (data not shown). After shifting to 36°C for 2 hours, sla2Δ cells became round and swollen and the actin patches became completely delocalized around the cortex, and medial rings could no longer be detected (Table 2). sla2Δ cells with extreme polarity defects had microtubules curling under the cortex (data not shown). However, at 25°C cells that still retained polarity showed a normal microtubule cytoskeleton, with three to four bundles running along the main axis of the cell (Fig. 1B). In these cells, Tea1 accumulated at opposite ends, as in wild-type cells (Fig. 1D).

To investigate further the role of End4/Sla2 in actin organization, we generated a talin deletion of End4/Sla2. As the talin-like domain is dispensable for most Sc-SLA2 functions in S. cerevisiae (Wesp et al., 1997), we reasoned that this deletion might specifically interfere with actin regulation. A strain bearing a talin-truncated version of End4/Sla2 (sla2Δtalin) was viable at all temperatures and grew at rates similar to those of wild type. As in budding yeast (Wesp et al., 1997; Yang et al., 1999), no obvious difference in bulk lipid endocytosis (by up-take of the lipophilic steryl dye FM4-64) was detected between wild-type and sla2Δtalin cells (data not shown). As shown by calcofluor staining (Fig. 2B), the only defect detected in this strain was that most cells grew from only one end. Like wild-type cells, sla2Δtalin cells had actin cables running along the main axis of the cell and cortical actin patches at growing ends. Unlike the wild-type population, where only 15% of cells were in a pre-NETO state and showed monopolar actin, 78% of sla2Δtalin cells had a monopolar distribution of cortical actin patches (Fig. 2A,B). The growth pattern of sla2Δtalin cells was further analysed by time-lapse phase-contrast microscopy. In wild-type cells after fission, both daughter cells activate growth from the old end (growing in the previous cycle). In contrast, in 53% of sla2Δtalin cells one daughter cell grew from the old end and one from the new end (Fig. 2D).

Deletion of the talin-like domain did not affect End4/Sla2 protein levels (Fig. 2C). Moreover, sla2Δtalin cells showed no obvious defect in bulk lipid endocytosis, using the FM4-64 uptake assay (data not shown). Thus as in S. cerevisiae, the C-terminal talin-like domain is dispensable for most End4/Sla2 functions. However, in S. pombe the End4/Sla2 talin-like domain is required for activation of growth at the new end, and possibly mediates recruitment of actin to the new end.

Further support for this was obtained by mutating arginine 1083, which abolishes binding of the isolated I/LWEQ-module to F-actin (McCann and Craig, 1999). Cells bearing this point mutation (Sla2R1083G) in End4/Sla2 showed NETO defects; 62% of cells grew in a monopolar fashion with a monopolar actin pattern (Fig. 2A).

To test if the talin-like domain plays an essential role in the switch to bipolar growth, we checked whether it was possible to promote NETO in sla2Δtalin cells by making actin monomers available in the cell. A pool of free actin monomers could be generated with a pulse of latranculin A (latA) (Rupes et al., 1999). Cdc10-129 and cdc10-129 sla2Δtalin cells were blocked for 2 hours at restrictive temperature (36°C). After the block, cells of both genotypes showed a monopolar actin distribution: 89% in cdc10-129 and 98% in cdc10-129 sla2Δtalin cells (Fig. 3A). Cells were than treated for 6 minutes with 40 μM latA, which completely disrupted actin localization (Fig. 3B). After washing out the drug, actin localization was followed for up to 120 minutes at 36°C. Within 1 hour most cdc10-129 cells showed a bipolar actin distribution, whereas most cells bearing the talin deletion in End4/Sla2 had monopolar actin patches. As shown in Fig. 3A, 2 hours after wash out of the drug, 76% of cdc10-129 cells showed a bipolar actin distribution compared to 22% of cdc10-129 sla2Δtalin cells. These results suggest that End4/Sla2 talin-like domain plays an essential role in the switch from monopolar to bipolar growth and in the organization of cortical actin patches at the new end.

We next determined the localization of End4/Sla2 in the cell throughout the cell cycle by using a C-terminal Sla2-GFP fusion expressed from the sla2 endogenous promoter. The GFP fusion protein is functional, as Sla2-GFP fully rescued the defects observed in sla2Δ cells. During interphase, Sla2-GFP accumulated in cortical dots at both cell ends (Fig. 4A) at the site of cell growth and actin patch accumulation. Evaluation of individual Sla2-GFP patches revealed a lifetime of 110-150 seconds. During early mitosis, Sla2-GFP dots form a ring in the middle of the cell, overlying the nucleus, which persists until the end of mitosis (Fig. 4A). Upon cell division, Sla2-GFP localization was re-established at the old end. Time-lapse movies (data not shown) of single cells expressing Sla2-GFP showed that after cell division only a few dots were left at the new end. Shortly after division, Sla2-GFP quickly accumulated again at cell ends, giving rise to the bipolar interphase pattern (data not shown). To determine whether End4/Sla2 accumulates only at growing cell ends, we examined the pattern of Sla2-GFP localization in the temperature-sensitive cell cycle mutants cdc10-129 and cdc25-22 (Mitchison and Nurse, 1985), which block before and after activation of bipolar growth, respectively. When shifted to the restrictive temperature, cdc10-129 cells blocked, before NETO, with monopolar actin and monopolar Sla2-GFP distribution (Fig. 4D). Conversely, cdc25-22 cells, at restrictive temperature, blocked entry into mitosis and accumulated as long rod-shaped cells with bipolar actin and bipolar Sla2-GFP (data not shown). We conclude that End4/Sla2 has a distribution that mirrors that of the cortical actin patches, accumulating at actively growing tips during interphase and re-localizing to a central ring prior to mitosis. However, co-staining of Sla2-GFP and actin showed only a partial colocalization of End4/Sla2 dots and actin patches (Fig. 4D,E).

The proper location of new growth zones relies on the polarity marker Tea1, which defines cell geometry by marking sites of cell growth (Mata and Nurse, 1997). Double labelling of Sla2-GFP-expressing cells with GFP and Tea1 antibodies showed some colocalization of the two proteins at cell tips (Fig. 4F). As End4/Sla2 also associates closely with the actin cytoskeleton and the growth machinery, we tested whether its localization depends on Tea1. The central accumulation of Sla2-GFP at the ring was unaffected in tea1Δ cells, whereas the end localization of Sla2-GFP was abnormal. Instead of the normal bipolar localization, most cells showed monopolar Sla2-GFP accumulation at the growing end (Fig. 4B) throughout the cell cycle, suggesting a role for Tea1 in proper End4/Sla2 localization. However, because Tea1 influences microtubular organization, the aberrant pattern of Sla2-GFP localization could also be an indirect consequence of the defective microtubular array observed in tea1Δ cells. To distinguish between these two possibilities, we examined the pattern of Sla2-GFP localization in tea1Δ(200) cells, which show a wild-type array of microtubules but fail to accumulate the truncated version of Tea1 at cell ends (Behrens and Nurse, 2002). tea1Δ(200) cells showed monopolar Sla2-GFP accumulation which occurred only at growing ends, even though microtubules were normal in these cells (Fig. 4C). Hence, Tea1 is necessary to recruit End4/Sla2 to the new end where growth has not previously occurred, independently of its effect on microtubules.

To further test if the accumulation of End4/Sla2 at cell ends required the microtubule or actin cytoskeletons, we performed a cdc10 block and release experiment in the presence of 25 μg/ml MBC, which depolymerizes microtubules, or 40 μM latA, which sequesters actin monomers leading to complete disassembly of the actin cytoskeleton. Cells were blocked at the restrictive temperature of 36°C for 3 hours to allow most cells to accumulate in G1 with monopolar Sla2-GFP. Cells were then released to the permissive temperature in MBC or latA to allow cells to proceed through the cell cycle in the absence of microtubules or F-actin structures, respectively. As shown in Fig. 5, neither the MBC nor the latA treatment impaired Sla2-GFP accumulation at the new end. Bipolar accumulation of Sla2-GFP was also observed when cells were treated with both MBC and latA together, excluding the possibility of a redundant dependence between the two systems (Fig. 5). Taken together these observations suggest that End4/Sla2 end localization does not directly require the actin and microtubular cytoskeletons.

In fission yeast, membrane trafficking is required for proper actin ring dynamics and affects its localization (Pardo and Nurse, 2003). The presence of a ENTH domain, shown to be involved in membrane binding (Ford et al., 2001; Itoh et al., 2001) at the C terminus of End4/Sla2, and the tight association of End4/Sla2 with the cortical membrane after latA treatment, led us to investigate whether membrane trafficking is also involved in localizing End4/Sla2 at cell ends.

Brefeldin A (BfA) blocks protein secretion by blocking vesicular transfer from the endoplasmic reticulum to the Golgi and therefore disrupting the Golgi apparatus. Treatment with BfA does not affect Tea1 localization (Mata and Nurse, 1997), but it caused Sla2-GFP to disappear completely from cell tips, becoming randomly distributed throughout the cell within 5 minutes (Fig. 6Ac), suggesting that End4/Sla2 accumulation at cell ends might be mediated by active vesicular trafficking. Conversely, treatment with 40 μM latA (Fig. 6Ab) for up to 1 hour did not affect Sla2-GFP end localization. Similarly, deletion of the actin binding talin-like domain has no effect on End4/Sla2 localization. As shown in Fig. 6Ba, Sla2Δtalin-GFP accumulates at both cell ends. Co-staining of Sla2Δtalin-GFP and actin (Fig. 6C) shows accumulation of Sla2Δtalin-GFP at both growing and not-growing ends.

However, similarly to budding yeast where latA treatment abolishes Sc-SLA2 movement (Kaksonen et al., 2003), the rapid and apparently random movement of Sla2-GFP dots near the cell cortex was completely blocked by addition of 40 μM latA to the medium (Fig. 6D). BfA-dependent delocalization of Sla2-GFP was also suppressed by the simultaneous addition of 40 μM latA (Fig. 6Ad), as well as by the deletion of the talin-like domain (Fig. 6Bc). Thus End4/Sla2 localization requires active vesicular trafficking and is only maintained at cell ends for a short time of around 5 minutes. We conclude that rapid removal of End4/Sla2 from cell ends requires the actin cytoskeleton and the talin-like domain.

We have identified End4/Sla2 as a protein required to recruit actin and the growth machinery to the new cell end in fission yeast. Phenotypic analysis of sla2Δ mutants showed that this gene is required for proper actin organization and cell morphogenesis. Cells lacking End4/Sla2 have an aberrant shape and a depolarized cortical actin cytoskeleton. This phenotype is reminiscent of that shown by deletion of Sc-sla2. In budding yeast, sla2Δ cells are spherical and have delocalized cortical actin structures, with aggregated actin patches dispersed throughout the cytoplasm (Holtzman et al., 1993).

Deletion of Sc-sla2 also caused defects in endocytosis (Wesp et al., 1997). Sc-SLA2 is involved in the internalization step of endocytosis (Kaksonen et al., 2003; Raths et al., 1993), but does not require its C-terminal talin-like domain for bulk lipid endocytosis (Baggett et al., 2003). Recent work by Iwaki et al. showed that in fission yeast, End4/Sla2 is also required for endocytosis (Iwaki et al., 2004). Deletion of the end4/sla2 gene inhibits delivery of FM4-64 to the vacuolar membrane, accumulation of luciferase yellow CH and internalization of the membrane protein Map3 (Iwaki et al., 2004). Although we cannot exclude a defect in receptor-mediated endocytosis as is observed in S. cerevisiae at elevated temperatures (Baggett et al., 2003), we observed no defects in fluid phase endocytosis (assayed by FM4-64 uptake) when the C-terminal talin-like domain of End4/Sla2 was deleted. Our analysis of the sla2Δtalin strain instead uncovered a novel function for End4/Sla2 in establishment of new growth zones in fission yeast. Moreover, in contrast with previous work in S. cerevisiae (Yang et al., 1999) and in Candida albicans (Asleson et al., 2001), which attributed only a marginal role to the talin-like domain, our data suggest a role for the talin-like domain in the recruitment of actin and activation of growth at new cell ends. Sla2Δtalin cells are, in fact, NETO defective, with monopolar actin patches. End4/Sla2 talin-like domain mutant cells fail to activate growth at the new end but do polarize growth at the old end and grow in a monopolar fashion. Unlike ssp1Δ cells, which can activate bipolar growth upon latA treatment (Rupes et al., 1999), sla2Δtalin cells have lost their intrinsic ability to activate growth at the new end, and their NETO defect cannot be overcome by an artificial increase in free actin monomers. Hence End4/Sla2 has an essential role in linking actin organization and growth. Since the truncated protein localizes to both the growing and the non-growing end, however, the role played by the talin-like domain might be restricted to establishment of bipolar growth.

As the talin-like domain binds actin both in vitro (McCann and Craig, 1999) and in two-hybrid assays (Yang et al., 1999), it is tempting to suggest that End4/Sla2 acts downstream of the polarity machinery at the interface with actin, and that the NETO defect seen in sla2Δtalin cells is due to the inability of the truncated protein to either polymerize or stabilize actin at the new end despite the presence of polarity markers. However, given that it has also been suggested that the talin-like domain is required for protein-protein interaction and regulation of Sc-SLA2 activity (Yang et al., 1999), an alternative explanation could be that the truncated form is inactive or unable to interact with the growth machinery.

Full deletion of Sla2/End4, which affects endocytosis as well as actin organization, leads to complete loss of polarity, whereas deletion of the talin-like domain, which does not interfere with bulk endocytosis, does not affect polarized growth but does influence establishment of new sites of polarized growth. Interestingly in budding yeast, sla2-6 mutant cells, which, unlike sla2Δ cells, grow well at all temperatures and show polarized actin, exhibit a bipolar budding defect (Yang et al., 1997). This suggests that in both fission and budding yeast Sla2 might have two independent functions: to maintain polarized cell growth (requiring the full length protein) and to promote NETO/distal budding. Formally, however it is also possible, at least in S. pombe, that End4/Sla2 is required only for establishment of polarized growth, with activation at the new end requiring a higher level of activity and therefore being more readily affected by alteration of the protein, such as the deletion of the talin-like domain. As previously suggested for Orb3, Orb8 and Orb9 (Snell and Nurse, 1994; Verde et al., 1995), End4/Sla2 might be required to reorganize the actin cytoskeleton after a major cell cycle transition, such as mitosis. Remarkably the general loss of polarization observed in sla2Δ cells resembles that shown by orb mutants at restrictive temperature (Snell and Nurse, 1994; Verde et al., 1995). Interestingly most of the orb genes identified so far encode protein kinases and Sc-SLA2 is phosphorylated (Ficarro et al., 2002) in logarithmically growing cells. This phosphorylation has been suggested as a potential regulatory switch to trigger End4/Sla2-actin interaction at the cell cortex (Yang et al., 1999) making End4/Sla2 a candidate downstream target of the Orb morphogenetic pathway.

In the rod-shaped fission yeast it is thought that the plus ends of microtubules define cell ends by delivering polarity markers such as Tea1 (Mata and Nurse, 1997). End4/Sla2 in also enriched at cell ends during interphase and is therefore a good candidate to play a direct role in the recruitment/assembly of cortical polar actin. Consistently, End4/Sla2 end localization is independent of microtubules but requires Tea1 to recognize the end where growth has not previously occurred. Cells lacking Tea1 do not accumulate End4/Sla2 at the new end and fail to undergo NETO. Therefore our work places End4/Sla2 as a link between microtubules and the machinery involved in sensing overall cell space and the actin cytoskeleton, which leads to polarized growth. In this model Tea1 would mark new sites of cell growth and recruit End4/Sla2, which in turn could initiate accumulation of cortical actin.

Tea1 and End4/Sla2 only partially colocalize at cell ends. However, if Tea1 is only required for the initial recruitment of End4/Sla2 but not for its maintenance at cell ends, their interaction could be only transient, and no longer be necessary once the End4/Sla2-cortex interaction was established. The functional interaction between Tea1 and End4/Sla2 might reflect a direct physical interaction, and consistent with this we could detect a direct interaction between the C terminus of Tea1 and End4/Sla2 in a two-hybrid assay (our unpublished observation), although we could not confirm it in vitro in fission yeast cell extracts. It is also possible that End4/Sla2 acts in conjunction with other factors such as Bud6, a factor required for NETO and shown to physically associate with Tea1 (Glynn et al., 2001). Interestingly, bud6Δ and sla2Δtalin show the same growth patter (Glynn et al., 2001). Moreover, the proper localization of both Bud6 (Jin and Amberg, 2000) and End4/Sla2 (BfA experiment, this study) is mediated by the secretory pathway and requires Tea1. Interestingly, it was postulated that the secretion and endocytosis might be coupled: the secretory pathway being required to replenish components needed at the plasma membrane for endocytosis (Riezman, 1985). However, S. cerevisiae end3 mutant cells, which are defective in endocytosis, still grow and therefore have an active secretion pathway (Raths et al., 1993). Although the coupling might not be essential, a functional link might still exist. Localized secretion may be required to accumulate proteins at sites of growth, and localised endocytosis could define and delimit the growth zone to cell tips by removing excess factors from neighbouring areas. In this scenario, depolarized endocytosis would result in loss of cellular domains and consequently in loss of polarity, as seen in sla2Δ cells.

Studies in other organisms have shown that, as in fission yeast, Sla2 homologues are enriched in subcellular regions, which show elevated membrane turnover and actin accumulation. Hip1, the human homologue of Sla2, colocalizes with and binds to clathrin and adaptor protein-2 (AP2) (Engqvist-Goldstein et al., 2001; Mishra et al., 2001), both highly enriched at presynaptic nerve terminals, where they facilitate rapid retrieval of synaptic vesicles (De Camilli et al., 2001; Sun et al., 2002). Hip1 was identified through its interaction with huntingtin (Kalchman et al., 1997; Wanker et al., 1997), a protein containing a polyglutamine stretch. Polyglutamine expansion causes a decrease in the Hip1-huntingtin interaction and is associated with the onset of Huntington's disease (HD). Huntingtin, like Tea1, associates with microtubules (Gutekunst et al., 1995; Hoffner et al., 2002; Muchowski et al., 2002; Tukamoto et al., 1997), and so Sla2/Hip1 family proteins may have a conserved role working at the interface between the actin and microtubular networks.

We thank all members of the Nurse lab for help and discussion. Special thanks to J. Hayles, K. Leonhard, R. Carazo-Salas and M. Pardo for valuable comments on the manuscript. S.C. was supported by a Marie Curie postdoctoral fellowship and by a CRUK postdoctoral fellowship. R.B. was supported by ICRF and by a Boehringer Ingelheim Fonds fellowship.