Growth-speed-correlated localization of exocyst and polarisome components in growth zones of Ashbya gossypii hyphal tips Free

Filamentous fungi comprise a large group of very diverse species. Some are beneficial, others cause devastating plant diseases or are becoming increasingly notorious as human pathogens (Harris et al., 2005). They grow as multinucleated tubular cells called hyphae by polar surface expansion at their tips (Bartnicki-Garcia and Lippman, 1969). Expansion of the plasma membrane and the fungal cell wall requires continuous incorporation of lipids and cell wall components, which are synthesized throughout the hyphae before being transported to the tip by secretory vesicles. In some species, hyphal surface expansion rates can amount to 100 μm²/minute, much higher than the well-studied growth of yeast buds, which expand at around 1 μm²/minute (Collinge and Trinci, 1974; Karpova et al., 2000; Park and Bi, 2007).

One hallmark of fungal polar growth is the persistent tip localization of polarity markers that, in budding yeast, only transiently localize to tips of daughter cells. Examples are the formins SepA and Bni1 (Ozaki-Kuroda et al., 2001; Schmitz et al., 2006; Sharpless and Harris, 2002), the p21 activated protein kinase Cla4 (Ayad-Durieux et al., 2000; Holly and Blumer, 1999) and the polarisome component Spa2 (Crampin et al., 2005; Knechtle et al., 2003; Snyder, 1989; Virag and Harris, 2006). The mechanisms leading to the differences between stable and transient localization of polarity factors are not known. However, it has been noted that cortical actin patches do not localize to the entire hyphal tip, leaving at its front a confined space for polar growth components (Knechtle et al., 2003; Taheri-Talesh et al., 2008).

Another hallmark of hyphal growth is the tip-located Spitzenkörper (German for `tip body'), a complex multicomponent structure dominated by vesicles (reviewed by Harris et al., 2005; Steinberg, 2007). Computer simulations suggest that it serves as vesicle supply centre at hyphal tips (Bartnicki-Garcia et al., 1989; Gierz and Bartnicki-Garcia, 2001). Proteins that are commonly associated with secretory vesicles are found in the Spitzenkörper like a v-SNARE in Aspergillus nidulans (Taheri-Talesh et al., 2008) or cell wall synthesizing enzymes in Neurospora crassa (Riquelme et al., 2007). In addition, the GTP-binding protein Sec4, a regulator of vesicle transport and fusion, takes a Spitzenkörper-like shape in hyphal tips of Ashbya gossypii (Schmitz et al., 2006). Furthermore, filamentous actin (f-actin) is required for the Spitzenkörper as the Spitzenkörper is sensitive to disruption of the actin cytoskeleton (Crampin et al., 2005; Taheri-Talesh et al., 2008) and as formins, which catalyze actin cable polymerization, are found in the Spitzenkörper of different species (Crampin et al., 2005; Harris et al., 2005).

We study hyphal growth of the cotton pathogen A. gossypii (Ashby and Nowell, 1926). Its genome shows 92% gene order conservation with the Saccharomyces cerevisiae genome (Dietrich et al., 2004). Analogous to S. cerevisiae, secretory vesicles are believed to arrive at hyphal tips via myosin-dependent transport along actin cables. These vesicles are tethered to the plasma membrane by the exocyst and fuse. The exocyst is a conserved complex that consists of eight components in budding yeast: ScExo70, ScExo84, ScSec3, ScSec5, ScSec6, ScSec8, ScSec10 and ScSec15. The exocyst was shown to be essential for fusion of secretory vesicles and is found at sites of exocytosis in S. cerevisiae daughter cells and the forming septum (Finger et al., 1998; Guo et al., 1999; TerBush et al., 1996). It interacts with factors that control polar growth such as ScCdc42, ScRho1, ScRho3 and ScBem1, and with proteins of the late secretory system (Adamo et al., 1999; France et al., 2006; Guo et al., 2001; Zhang et al., 2001). Therefore, it constitutes an interface between cell polarity and the secretory pathway (Guo et al., 1999). Although the importance of polarized exocytosis in hyphal growth is undisputed, no systematic investigations have been performed in filamentous fungi. Only recently, it was described that the A. nidulans Sec3 homolog localizes to a narrow area at the hyphal tip (Taheri-Talesh et al., 2008) and that deletion of CaSEC3 blocks formation of hyphae in Candida albicans (Li et al., 2007).

In order to understand hyphal morphogenesis, it is not only important to know the contribution of single factors but also how factors work together to form the machinery that governs polar growth. Therefore, we have begun to analyze the exocyst, cell polarity determinants and the polarisome in A. gossypii hyphae that elongate with different speeds. Mature hyphae grow up to 40 times faster than newly established hyphae, i.e. germ tubes and emerging lateral branches (Ayad-Durieux et al., 2000; Knechtle et al., 2003). We hypothesize that changes in growth speed would influence the localization or dynamics of components in the tip.

First, we report on light and electron microscopy experiments to search for a Spitzenkörper in hyphal tips of A. gossypii. Then we address the issues of whether all exocyst components are essential in A. gossypii, which area of the tip is associated with exocyst components and whether this area increases during acceleration of hyphal growth. We also asked whether key polarity factors and polarisome components localize to the exocyst area, whether the localization of these factors changes with increasing elongation speed, whether the localization of polarisome components depend on each other, and whether zones of exocytosis and endocytosis overlap or not. Finally, we test the role of the cytoskeleton for the distribution of exocyst and polarisome components.

Hyphal elongation speeds increase during the development of A. gossypii. Emerging hyphae of germlings elongate at about 0.2 μm/minute. Within 24 hours, hyphal growth speed accelerates to 3.5 μm/minute. Examples of slow and fast hyphae are shown in Fig. 1A. Interestingly, the diameter of hyphae changes with growth speed. Faster growth correlates with a hyphal diameter increase from around 3.5 to 5.0 μm (Fig. 1B). This implies that a doubling of growth speed leads to a more than twofold increase in surface expansion rate and explains the non-linear relationship between hyphal speed and surface expansion rate (Fig. 1C). New membrane and cell wall material is transported to sites of growth via secretory vesicles. The demand for delivery and processing of these vesicles in hyphal tips increases more than tenfold during acceleration of growth. Does this lead to other observable changes in addition to the described increase in hyphal diameters? If growth of fast hyphae is monitored by phase-contrast light microscopy on Ashbya full medium (AFM) solidified with gelatine, a dark apical body is observed in the tip dome (Fig. 1D). As expected, this area shows a strong accumulation of vesicles when analyzed by transmission electron microscopy (Fig. 1E). Two major classes of vesicles were observed with diameters of 20-40 μm and 80-100 μm, respectively (Fig. 1F). This corresponds well to the vesicle sizes reported for other filamentous fungi (Harris et al., 2005; Steinberg, 2007). None of the inspected thin sections showed a fusion intermediate of a vesicle with the plasma membrane, most probably because vesicle fusions proceed very fast. Taken together, these data indicate that fast growing A. gossypii hyphae possess a Spitzenkörper. In slow hyphae, Spitzenkörper were not observed (as will be shown later).

Acceleration of hyphal growth and the concomitant increase of the surface expansion rate are only possible with an increased fusion rate of secretory vesicles. First, we wanted to know where these vesicles fuse with the plasma membrane, i.e. over the entire surface or only in a restricted area of the tip, and whether this area would enlarge with an increased demand for vesicle fusion. To address these issues, we analyzed the exocyst of A. gossypii. Homologs of the eight S. cerevisiae exocyst genes are present in the A. gossypii genome (Table 1). Individual deletions of the eight A. gossypii exocyst genes were lethal (Table 1), which is in agreement with the essential function of exocytosis in hyphal growth. We assessed the localization of GFP and YFP fusions to three exocyst components, AgExo70, AgSec3 and AgSec5 in hyphae elongating with different speeds. In slowly growing hyphae AgExo70-GFP localized as a cortical cap in the tip dome. At higher elongation speeds, it accumulated inside the hyphal tip finally taking a spheroid shape (Fig. 2A). For quantification, the expansion of the AgExo70-GFP signal parallel to the hyphal growth axis was plotted against the growth speed (Fig. 2B). The data points show that the size of the AgExo70-GFP signal gradually enlarges with increasing growth speed. Hyphal tips with cortical AgExo70-GFP caps elongated with less than 1.0 μm/minute. Tips with a spheroid AgExo70-GFP distribution extending about 1.2 μm along the hyphal axis grew faster than 1.4 μm/minute. Crescent-like localization were observed in hyphae with intermediate growth speeds (see examples marked a-c in Fig. 2A,B). Very similar localization, cortical caps at low speed and enlarging distributions up to spheroids with increasing growth speeds were observed for AgSec3-YFP (Fig. 2C,D) and AgSec5-GFP (data not shown). Thus, the localization of AgExo70-GFP, AgSec3-YFP and AgSec5-GFP in hyphal tips correlated closely with growth speed.

We next tested whether the tip-based spheroid-shaped distribution of exocyst components observed in fast hyphae correlates with the dimensions of the Spitzenkörper shown in Fig. 1D. The gelatine-enriched medium that allows resolution of intracellular details showed strong background fluorescence. Therefore, we used the lipophilic dye FM4-64, which stains the Spitzenkörper in different fungal species most probably owing to fast recycling of endocytosed membrane material in the tip region (Fischer-Parton et al., 2000). In A. gossypii, FM4-64 accumulation in the Spitzenkörper became apparent between 2.7 and 5.5 minutes after dye addition (n=9), whereas staining of putative vacuoles was visible only after prolonged incubations (see Fig. S1 in the supplementary material). Interestingly, fast but not slow A. gossypii hyphae accumulated FM4-64 in a spherical region in the tip (Fig. 3A,B). The spherical region stained with FM4-64 overlapped with the spherical localization of AgExo70-GFP, AgSec3-YFP (Fig. 3C) and AgSec5-GFP (data not shown). No FM4-64-stained Spitzenkörper could be detected at growth speeds at which exocyst components were exclusively observed as cortical caps (Fig. 2; Fig. 3B). Together, these observations strongly suggest that, in fast hyphae, exocyst components are part of the Spitzenkörper. Furthermore, not only Spitzenkörper localization of exocyst components but Spitzenkörper formation itself depends on hyphal growth speed.

We wondered whether the exocyst is active at the cortical cap and in the Spitzenkörper region. To answer this question, we localized A. gossypii homologs of budding yeast proteins, which are known to promote exocytosis by interacting with the exocyst. Among them were homologs of the conserved GTP-binding protein ScCdc42 and the adaptor protein ScBem1 (France et al., 2006; Zhang et al., 2001). A functional GFP-AgCDC42 fusion was obtained using a novel, selection-based screening approach (described in the Materials and Methods). As the exocyst interacts with Cdc42 in its GTP-bound active state (Zhang et al., 2001), we additionally localized AgCdc24, the putative guanine nucleotide exchange factor for AgCdc42 (Wendland and Philippsen, 2001). Strikingly, GFP-AgCdc42, AgCdc24-GFP and AgBem1-GFP localized as cortical caps only to the tip dome, even in fast hyphae that displayed an FM4-64-stained Spitzenkörper (Fig. 4). This strongly indicates that AgCdc42 interacts with the exocyst at the cortex, where exocytosis most likely takes place. The GFP-fluorescence in the Spitzenkörper probably represents exocyst components that are associated with vesicles, because in yeast, all the exocyst components, except Sec3, are associated with secretory vesicles (Boyd et al., 2004; Guo et al., 1999).

We examined tips of fast hyphae (extension rate >1.50 μm/minute) displaying a spheroid AgExo70-GFP distribution. Additionally to the Spitzenkörper-based AgExo70-GFP pool, a pronounced cortical AgExo70-GFP cap was observed (Fig. 5A). This cortex-associated AgExo70-GFP most probably indicates the zone where the exocyst promotes secretory vesicle fusion with the plasma membrane. As stated above, the increasing surface expansion rates that occur during hyphal acceleration demand an increase in fusion of secretory vesicles. A way to increase the fusion rate would be to expand the area where vesicle fusion happens. To test whether this is the case, we measured the area of the cortical AgExo70-GFP cap and compared it with the hyphal growth speed (Fig. 5B). Interestingly, the sizes of this area ranged from 5 to 15 μm², without any clear indication of a much larger cortical exocyst area at higher growth speeds. The greater than tenfold increase in surface expansion rate that is observed during hyphal acceleration (Fig. 1C) is thus not mediated by a substantial expansion of the apical vesicle fusion zone. Consequently, compared with slow hyphae, exocytosis has to be much more efficient in a restricted area at the cortex of fast growing hyphal tips. Though no correlation with growth speed was found, the cortical area of AgExo70-GFP enrichment was quite variable. We thus wanted to test whether this variability reflects normal fluctuations over short times or whether hyphae display slowly changing apical vesicle fusion zones of different sizes. We monitored the AgExo70-GFP distribution in time-lapse movies (Fig. 5C; see Movie 1 in the supplementary material). Within the 60 seconds that are covered by this movie, the apical vesicle fusion zone enlarged from 10 μm² to over 15 μm² and shrunk again to 10 μm² (Fig. 5D). During this time period, hyphal growth speed did not change (Fig. 5D). The AgExo70-GFP marked Spitzenkörper stayed at its position in the hyphal apex, although its shape was changing constantly (Movie 1). Even though, size and position of the vesicle fusion zone as well as the shape of the Spitzenkörper fluctuate, on average they are centered on the growth axis (Fig. 5E).

Another complex involved in polar surface expansion is the polarisome, a poorly defined network of interacting proteins that was mainly studied in budding yeast (Park and Bi, 2007). It consists of ScSpa2, ScPea2 and ScBud6 (Sheu et al., 1998). As ScSpa2 and ScBud6 are involved in the localization and regulation of the formin ScBni1, the latter is often considered to be a fourth member of the polarisome (Evangelista et al., 1997; Fujiwara et al., 1998; Moseley et al., 2004). The A. gossypii genome encodes homologs for all four proteins. Loss of the formin AgBni1, which mediates the formation of the actin cables, is lethal (Schmitz et al., 2006). Deletion mutants of AgSPA2 are viable but hyphae elongate with decreased speed (Knechtle et al., 2003). Similarly, we found reduced growth rates for Agpea2Δ and Agbud6Δ strains. Radial growth speeds were 1.1 μm/minute, about one-third of wild-type mycelium on AFM at 30°C (Fig. 6A; Table 2). This similar reduction in maximal hyphal elongation indicates an important role for the polarisome components to reach fast growth. We wanted to test whether the role of the three non-essential polarisome proteins for fast growth is reflected by their tip localization. We also included AgBni1 in this analysis. In slow growing hyphae, fusions of polarisome components to GFP or YFP localized to the tip cortex (Fig. 6B,C), as was the case for the exocyst components or cell polarity factors. This finding suggests that all the polarisome components also function in hyphal growth at lower growth speeds. In fast hyphae, GFP-AgBni1, AgSpa2-GFP and AgPea2-YFP were observed in a spherical localization (Fig. 6B,C). Crescent-shaped localization were observed at intermediated growth speeds (Fig. 6B). FM4-64 staining revealed that AgSpa2-GFP, AgPea2-YFP and GFP-AgBni1 localize to the Spitzenkörper (Fig. 6D). By contrast, GFP-AgBud6 was restricted to the cortex independent of growth speed (Fig. 6B,C) even though an FM4-64-stained Spitzenkörper was observed (Fig. 6D). The localization of the polarisome components overlap only at the cortex of the hyphal tip. Therefore, a hypothetical protein complex containing all four polarisome components could exist only at the cortex.

In fast hyphae GFP-AgBni1 was not present in the entire Spitzenkörper like AgExo70-GFP and AgSpa2-GFP (Fig. 3C; Fig. 6D upper panels). We tested potential differences in the distribution of these proteins by employing three-dimensional deconvolution of image stacks of 32 planes each. For each strain, 30 hyphal tips growing faster than 1.50 μm/minute were analyzed. We found that the localization of GFP-AgBni1, AgSpa2-GFP and AgExo70-GFP differed from each other. In 71% of all cases, GFP-AgBni1 was enriched in a small spot in the tip dome (Fig. 6E, black arrowhead). In the remaining 29% GFP-AgBni1 either was homogenously distributed in the Spitzenkörper or localized in a large, fuzzy core region that was connected to the cortex (not shown). Furthermore, a zone of reduced GFP-AgBni1 fluorescence at the very tip proximal to the core-region was observed in 58% of all cases (Fig. 6E, grey arrowhead). Similar to GFP-AgBni1, AgSpa2-GFP was enriched in a small spot in 52% of all tips though the fluorescence intensity difference between the spot and the surrounding AgSpa2-GFP signal was less pronounced (Fig. 6F, arrowhead). In 34% of the hyphae, AgSpa2-GFP displayed a uniform Spitzenkörper-like localization (Fig. 6F, bottom row), the remaining 14% displayed irregular or crescent-like AgSpa2-GFP localization (not shown). Enrichment of AgExo70-GFP in a central spot was not observed (Fig. 5A).

In order to test whether the localization of polarisome components depend on each other, we deleted the three non-essential polarisome genes in the strains with fluorescently labeled polarisome components. The mutant strains displayed hyphal diameters significantly larger than the wild type (Fig. 6G; Table 2). Fluorescently labeled polarisome components still localized to hyphal tips in the deletion strains, except AgPea2-YFP in Agspa2Δ (Fig. 6G; see Fig. S2 in the supplementary material). In some deletions, the fluorescence intensity at the tip was weaker than in wild-type strains. This most probably indicates a diminished recruitment to the tip, as expression of the fluorescently labeled polarisome components was not affected by the deletions (see Fig. S2 in the supplementary material). In all deletion strains, GFP-AgBni1, AgSpa2-GFP and AgPea2-YFP were restricted to the cortex of the tip dome. This was not only due to the low growth speeds inflicted by the mutations, as polarisome components frequently showed crescent-shaped localization patterns in wild-type hyphae with comparable speeds (Fig. 6G; see Fig. S2 in the supplementary material). For quantification, the ratio between the maximal fluorescence value in the tip and 10 μm subapical to the tip was determined (Fig. 6H). In summary, deletion of either AgSPA2 or AgPEA2 resulted in very strong reduction or loss of the tip localization of GFP-AgBni1, AgSpa2-GFP or AgPea2-YFP. These findings are consistent with the localization data in such that all the factors found in the Spitzenkörper also strongly depend on each other for localization. Although the fluorescence reduction was less pronounced, the GFP-AgBud6 localization was affected in Agspa2Δ and Agpea2Δ strains, and, vice versa, the localization of AgSpa2-GFP and AgPea2-GFP was affected in the Agbud6Δ strain. Strikingly, the maximal GFP-AgBni1 fluorescence in the tip was not decreased in Agbud6Δ, indicating that AgBni1 does not depend on AgBud6 for its concentration at sites of polar growth. Similar dependencies between polarisome components are described in budding yeast. ScSpa2 and ScPea2 play an important role for recruitment of ScBni1, ScSpa2 and ScPea2 (Fujiwara et al., 1998; Ozaki-Kuroda et al., 2001; Sheu et al., 1998; Valtz and Herskowitz, 1996), whereas lack of ScBud6 only slightly disturbs ScBni1 localization (Jin and Amberg, 2000; Ozaki-Kuroda et al., 2001). Thus, in both A. gossypii and budding yeast, Bni1, Spa2 and Pea2 seem to form a functional unit. These proteins interact together in budding yeast, they localize to the Spitzenkörper in A. gossypii and loss of one of these proteins affects the localization of the other two in both S. cerevisiae and A. gossypii.

Formins and f-actin are found in the hyphal tip region of different fungi, including A. gossypii (Bourett and Howard, 1991; Crampin et al., 2005; Harris et al., 2005; Schmitz et al., 2006; Srijayanthi et al., 1996; Taheri-Talesh et al., 2008). This led to the speculation that formin-mediated actin cables are responsible for Spitzenkörper integrity and tip growth by clustering and redistributing secretory vesicles that are transported on microtubules to the tip region (Harris et al., 2005). To test whether this model is applicable to A. gossypii, we treated growing hyphae with either nocodazole or latrunculin A to disrupt microtubules or f-actin, respectively. Our results confirmed findings that A. gossypii hyphae continue to elongate in the presence of nocodazole, indicating that efficient transport of secretory vesicles from subapical regions does not depend on cytoplasmic microtubules (Fig. 7A) (Gladfelter et al., 2006). On the other hand, disruption of the actin cytoskeleton leads to swelling of hyphal tips and lysis within 15 minutes (Fig. 7B) (Knechtle et al., 2006), confirming the essential role of F-actin in tip growth of filamentous fungi (Akashi et al., 1994; Torralba et al., 1998).

It has been reported that disruption of microtubules leads to reduced hyphal elongation rates (Fuchs et al., 2005; Horio and Oakley, 2005). Therefore, microtubules may be important for fast hyphal elongation in A. gossypii. To test this, the growth speed of nocodazole-treated and untreated hyphae was monitored in time-lapse movies (Fig. 7C). Nocodazole-treated hyphae extended with 2.7±0.2 μm/minute, which was similar to the mock-treated cells that grew with 2.9±0.2 μm/minute. A spherical AgExo70-GFP localization was still observed in 72% of hyphal tips 15 minutes after nocodazole administration (Fig. 7D), thus the Spitzenkörper-like distribution of AgExo70-GFP was only marginally influenced by the loss of microtubules.

In order to further characterize the role of actin in tip growth, we assessed the localization of AgExo70-GFP and AgSpa2-GFP in hyphae stained with Alexa568-phalloidin, which strongly labels the tip-enriched actin patches and, less intensely, actin cables (Fig. 7E,F). The staining shows a separation of the zone of exocytosis, indicated by AgExo70-GFP, and endocytosis, indicated by the dense area of actin patches (Huckaba et al., 2004; Kaksonen et al., 2003). It seems that the faster the hyphae elongate, the more the endocytic zone is shifted away from the tip (data not shown). Two to six faint actin cables could be observed subapically of the endocytic zone. Whether these cables extended into the tip or not could not be determined due the intense fluorescence of the actin patches. The integrity of the tip-localized AgExo70-GFP and AgSpa2-GFP pools were assayed by incubation of the hyphae with 200 μM latrunculin A. Only traces of cortical f-actin remained after 60 seconds and no f-actin staining was detected after 180 seconds of drug treatment. The spheroid-shaped AgExo70-GFP localization disappeared after 15 seconds in most of the hyphae, whereas the cortical AgExo70-GFP population persisted (Fig. 7G,H). The AgSpa2-GFP signal lost its spheroid shape after 15 seconds but two distinct areas of localization of AgSpa2-GFP were observed: a cortical cap and a dense spot that was observed in the proximity of the cortex (Fig. 7I,J). This spot may represent the dislocalized intense spot seen in the central Spitzenkörper region, potentially indicating a structural integrity of this spot. In a similar way, GFP-AgBni1 was found in a bright spot additional to the cortical cap after 60 seconds of latrunculin A treatment (data not shown). These observations suggest that cortex-associated pools of the investigated proteins are maintained in an actin-independent way, whereas the Spitzenkörper pool is very sensitive to perturbation of the actin cytoskeleton.

The fungus A. gossypii is a well-suited biological system in which to study sustained polar surface expansion. Its genome carries at syntenic positions the homologs for all S. cerevisiae genes that encode known polarity components. Despite this conservation at the genome level, A. gossypii exclusively grows in the form of fungal hyphae (sustained polar surface expansion), whereas S. cerevisiae proliferates by alternating cycles of bud growth (mainly non-polar surface expansion) and cell separation. The biosynthetic capacity for surface growth also differs between both organisms. The tip surface of A. gossypii hyphae can extend up to 50 times faster than the surface of a growing yeast bud. It is therefore not surprising that we found distinct differences associated with cell surface growth in both systems, even though the basic function of individual polarity factors, mainly known from studies in S. cerevisiae (Park and Bi, 2007), was maintained during evolution.

It is well established that many polarity factors only transiently localize to tips of growing buds of S. cerevisiae and that polarized localization is not maintained during the non-polar growth phase, as shown for ScBni1 (Ozaki-Kuroda et al., 2001). Our studies indicate an intimate association between permanent tip localization of polarity factors and sustained polar growth of A. gossypii hyphae. We further show that the shape of the tip localization correlates with growth speed. At slow speeds, the exocyst, cell polarity factors and polarisome components form a cap at the tip cortex (Fig. 8A). With increasing growth speeds, the exocyst and three of the four polarisome components gradually accumulated, additionally to the cortical cap, in the tip dome (Fig. 8B). Electron microscopy showed that vesicles accumulate in this region. A substantial part of these are secretory vesicles as GFP-AgSec4 assumes a spheroid-shaped localization in hyphal tips (Schmitz et al., 2006). It is also likely that endocytosis-derived vesicles participate in this vesicle pool as FM4-64, a lipid dye that is taken up via endocytosis, accumulates within a few minutes in tips of fast hyphae (see Fig. S1 in the supplementary material).

Our findings represent compelling evidence that fast, but not slow, A. gossypii hyphae form a Spitzenkörper. A vesicle reservoir is apparently needed in tips of fast hyphae to coordinate and satisfy the increased demand for vesicle fusion (Fig. 8C). The finding that a Spitzenkörper is not present at slow growth speeds in A. gossypii suggests that this structure is not necessary for hyphal growth per se. This is supported by observations in A. nidulans where Spitzenkörper localization of a v-SNARE was observed in mature hyphae but not in slowly growing germlings (Horio and Oakley, 2005; Taheri-Talesh et al., 2008). Furthermore, slow expansion rates may explain the absence of a Spitzenkörper in S. cerevisiae buds or in C. albicans buds and pseudohyphae (Crampin et al., 2005).

Our data indicate that the Spitzenkörper forms gradually when hyphal growth speed accelerates. At intermediate growth speeds, we observed crescent-like localization of exocyst or polarisome components that seem to represent intermediate states between cortical and Spitzenkörper localization. It is obvious that the output of the secretory pathway has to increase to reach faster hyphal elongation rates. A more active secretory system may lead to gradual accumulation of vesicles in the tip. Along with the increase in local vesicle concentration, the efficiency of vesicle fusion with the plasma membrane will increase until a new steady state between vesicle transport and consumption is reached (Fig. 8C).

Even the fastest hyphae seem to carry an excess of vesicles in their tips, thus most probably excluding subapical vesicle supply as control for maximal growth speed. It is conceivable that maximal polar surface expansion depends on the capacity for docking and fusion of secretory vesicles at the tip cortex. The cortical vesicle fusion zone is defined by the localization of the exocyst and its activating factors. The size of this zone varied among hyphae of similar speed owing to its dynamic nature. Time-lapse movies showed that it was restricted to the very tip with fluctuating rims (Fig. 5). Remarkably, the average cortical area for vesicle fusion increases only slightly during the observed tenfold increase in surface expansion. Thus, vesicle fusion is spatially restricted. When the maximal fusion rate is reached in the confined tip area, hyphal growth speed may not increase further.

In A. gossypii, actin patches, which most probably represent sites of endocytosis, are excluded from the exocytic zone at the very tip. By contrast, they are evenly distributed at the cortex of growing yeast buds (Kaksonen et al., 2003). As long as buds are expanding, one has to assume that sites of exocytosis and endocytosis co-exist at the bud surface. In hyphal tips of A. gossypii actin patches are absent from the tip zone where the polarisome localizes (Knechtle et al., 2003) and also from the zone of exocytosis in fast growing hyphae. Therefore, zones of exocytosis and endocytosis do not overlap in A. gossypii hyphal tips. In contrast to slow hyphae, the zone of endocytosis is shifted further away from the tip front in fast growing hyphae (Fig. 8D).

The proteins assessed in this study can be divided into two groups: proteins that localize to the cortex and the Spitzenkörper; and proteins that are restricted to the cortex. Exocyst components as well as the polarisome components AgSpa2, AgPea2 and AgBni1 accumulated in the Spitzenkörper, which mainly consists of vesicles. This localization pattern was not surprising for the exocyst components AgExo70 and AgSec5 as their budding yeast homologs are transported with secretory vesicles. Furthermore, there is also evidence that ScSpa2 associates with secretory vesicles (Shih et al., 2005), which could explain the localization of AgSpa2 to the Spitzenkörper. However, no association between ScPea2 and ScSec3 with vesicles was observed in similar assays (Boyd et al., 2004; Shih et al., 2005). Thus, there are two alternative explanations for the Spitzenkörper localization of these proteins in A. gossypii. It is possible that AgPea2 and AgSec3 associate with an ill-defined matrix in the Spitzenkörper region, but not with vesicles. Alternatively, AgPea2 and AgSec3 may, unlike their homologs in yeast, associate with vesicles directly or indirectly in A. gossypii. Such an association may not be needed in the relatively small S. cerevisiae cells, but may be necessary for long-distance transport in A. gossypii hyphae.

Other factors were restricted to the cell cortex, among them AgCdc42 and AgBud6. Interestingly, ScCdc42 and ScBud6 seem to be associated with vesicles in budding yeast and their localization depends on a functional secretory pathway (Jin and Amberg, 2000; Wedlich-Soldner et al., 2003; Zajac et al., 2005). Consequently, one would expect to find these factors in the Spitzenkörper in A. gossypii, which is not the case. There are two non-exclusive explanations for this observation. First, it is not known whether different types of secretory vesicles defined by their cargo exist in A. gossypii or whether all vesicles destined for fusion with the tip plasma membrane accumulate in the Spitzenkörper. It is thus possible that vesicles transporting AgCdc42 or AgBud6 do not accumulate or are too rare to be detected. Second, both ScCdc42 and ScBud6 are able to localize to sites of polar growth independently of actin in yeast, which suggests alternative localization mechanisms that do not depend on targeted vesicle transport (Ayscough et al., 1997). Similar mechanisms could be responsible for localization of AgCdc42 and AgBud6 to the tip cortex of A. gossypii.

AgBni1 is concentrated at the tip cortex and in the centre of the Spitzenkörper. Cdc42 binds to and activates Bni1 both in S. cerevisiae and A. gossypii (Evangelista et al., 1997; Schmitz et al., 2006). Furthermore, the actin polymerization activity of ScBni1 is stimulated by binding of the conserved ScBud6 C-terminal half in budding yeast (Moseley and Goode, 2005; Moseley et al., 2004). Therefore, two activators of AgBni1, AgCdc42 and AgBud6, are restricted to the cortex, while AgBni1 itself is also found in the center of the Spitzenkörper where no homolog of known AgBni1 activators was enriched. Interestingly, also in budding yeast, ScBni1 localizes to sites of polar growth, together with Rho-type GTP-binding proteins and ScBud6, and to sites in the cytoplasm. It was found that ScBni1 dynamically localizes to the bud tip, where it mediates formation of actin cables. At the same time, it is incorporated in actin filaments and redispersed by the continuous retrograde actin flow, leading to the observed cytoplasmatic ScBni1-speckles (Buttery et al., 2007). If we assume a similar mechanism in A. gossypii, simple clustering of cable-associated, inactive AgBni1 may lead to the observed AgBni1 concentration in the Spitzenkörper and, at the same time, may focus actin cables at this site.

As in A. gossypii, the A. nidulans formin SepA localizes to a defined spot in the Spitzenkörper (Harris et al., 2005). Furthermore, distinct core regions are observed in the Spitzenkörper of other fungi (Grove and Bracker, 1970; Lopez-Franco and Bracker Charles, 1996). These findings suggest that the Spitzenkörper is not only a plain vesicle accumulation but shows some degree of organization. Interestingly, the concept of a polarized vesicle-based structure that is involved in actin organization seems to be conserved even beyond fungi. In the green algae Chara globularis, actin filaments emanate from a Spitzenkörper that is located in the tips of rhizoids, which are tubular gravity-sensing cells (Braun et al., 1999; Braun et al., 2004). Spitzenkörper are thus present in a wide variety of organisms: in A. gossypii, which is closely related to budding yeast and which does not rely on microtubules for tip growth; in the hyphal form of the dimorphic fungus C. albicans (Crampin et al., 2005); in many filamentous ascomycete and basidiomycetes; and even in filamentous plant cells. This argues that, although it may have different evolutionary origins in these organisms, the Spitzenkörper constitutes a very successful adaptation to the demands of fast filamentous growth of walled cells.

A. gossypii media, culturing and transformation protocols are described by Ayad-Durieux et al. and by Wendland et al. (Ayad-Durieux et al., 2000; Wendland et al., 2000).

The actin cytoskeleton was stained (see Knechtle et al., 2003). Anti-tubulin immunofluorescence was carried out as described previously (see Gladfelter et al., 2006). A rat anti-tubulin antibody (YOL34; Serotec, Kidlington, UK) was used at a dilution of 1:50, AlexaFluor568 goat anti-rat (Invitrogen, Carlsbad CA, USA) at a dilution of 1:200. Nocodazole (Sigma-Aldrich, St Louis, MO), 15 μg/ml for liquid and 30 μg/ml for solid medium, was used to disrupt microtubules. The actin cytoskeleton was disrupted with final concentrations of 200 μM latrunculin A in liquid and 400 μM on solid medium. Control treatments were performed with DMSO.

All DNA manipulations were carried out according to Sambrook (Sambrook, 2001). The E. coli strain DH5alphaF' (Hanahan, 1983) was used as a host. PCR was performed using Taq DNA polymerase, the Expand High Fidelity PCR system or the Expand Long Template PCR system (Roche Diagnostics, Mannheim, Germany). Oligonucleotides (see Table S1 in the supplementary material) were synthesized either by MWG (Ebersberg, Germany) or Microsynth (Balgach, Switzerland). Plasmids were constructed as described in Table S2 in the supplementary material.

A. gossypii strains are listed in Table S3 in the supplementary material. Agleu2ΔAgthr4Δ (Altmann-Johl and Philippsen, 1996) was used for all transformations unless indicated otherwise and is referred to as wt. Homologous integration of the transforming DNA was verified by analytical PCR in the primary transformants (heterokaryotic, nuclei with different genetic configurations share a common cytoplasm) and in clonally purified strains (homokaryotic, nuclei are genetically identical).

Genes were deleted using the PCR-based one-step gene targeting approach with heterologous markers (see Wendland et al., 2000) or using an indirect plasmid-based approach. The name of the respective oligonucleotides, PCR templates and plasmids used to produce the deletion cassettes can be found in the A. gossypii strain table (see Table S3 in the supplementary material). The GEN3 cassette (Wendland et al., 2000) mediates resistance against G418, and the NAT1 cassette (D. Hoepfner, personal communication) mediates resistance against ClonNat (Werner Bioagents, Jena, Germany).

Fusions to GFP or YFP were accomplished by co-transformation of a PCR-generated cassette with an ARS-CEN vector containing the A. gossypii gene into the yeast strain DHD5 (Arvanitidis and Heinisch, 1994). The resulting plasmids were digested with the restriction enzymes indicated in Table S3 (see supplementary material) and used for transformation of A. gossypii. GFP-fusion constructs were integrated into the genome replacing the wild-type genes. Expression was driven by the native promoter except for K47 and K52, where GFP-AgBNI1 and GFP-AgBUD6 are under control of the S. cerevisiae HIS3 promoter. Western blot analysis showed that the ScHIS3 driven expression of GFP-AgBNI1 (K47) resulted in five to ten times increased proteins levels (not shown). The presented results were obtained with K47, the localization patterns and speed dependences observed for the AgBNI1 promoter driven GFP-AgBNI1 expression (K46) were similar but signal strength, and thus image quality, was lower.

A GFP-AgCdc42 fusion constructed analogous to GFP-AgRho1a (Kohli et al., 2008) was not functional. Therefore, a random linker library was inserted between GFP and AgCDC42 on pK12 by homologous recombination in the yeast strain DLY3067 (Moskow et al., 2000). The random linkers, which consisted of nine SNY-repeats (S: G or C, N: G,A,T or C, Y: T or C) and flanking regions were created in a PCR-based fill-in reaction from the oligonucleotides 05.301 and 05.302. Functional GFP-AgCDC42 fusion constructs were selected on glucose that shuts down production of ScCdc42 in DLY3067. A heterokaryotic Agcdc42Δ strain was transformed with the rescued pK49 library. Plasmids containing an ARS sequence from yeast can replicate in A. gossypii and are not integrated into the genome (Wright and Philippsen, 1991). Clonal purification resulted in A. gossypii strains that were homokaryotic for the genomic AgCDC42 deletion expressing GFP-AgCdc42 from a plasmid, which was verified by analytical PCR. Radial growth speed was estimated to select for fusion constructs that displayed maximal radial growth speeds. The random linker of the strain used for this study was coding for the peptide APPRRLVHP. Similar results were obtained with strains that differed in the random linker sequence (not shown).

The microscope set up is described by Knechtle et al. (Knechtle et al., 2003); the camera was a CoolSNAP HQ camera (Photometrics, Tucson AZ, USA). A 75 W XBO short arc lamp (Osram, Augsburg, Germany) or a Polychrome V monochromator (Till Photonics, Gräfelfing, Germany) served as illumination sources. Mycelium from the borders of 3-day-old A. gossypii colonies was inoculated on glass slides with a cavity (Roth, Reinach, Switzerland) filled with half-strength AFM agarose. 4 μl of 11 μM FM4-64 in AFM was applied directly on the sample if not mentioned otherwise. FM4-64-stained Spitzenkörper were observed after 10 minutes of incubation, stained samples were discarded 1 hour after dye addition. DIC images were processed using the `Unsharp Mask' feature from MetaMorph 6.2r6 (Molecular Devices, Downingtown, PA). Stacks with a z-distance between 0.3 and 0.8 μm were acquired and processed either with the `Remove Haze' function or the `Nearest Neighbour' tool of MetaMorph. Overlays were carried out with the `Overlay Images' function of MetaMorph. Stacks that were used for blind deconvolution with AutoDeblur 7 (MediaCybernetics, Silver Spring MD, USA) contained at least 32 image planes with a z-distance of maximally 0.3 μm. The fluorescence images shown are maximum or sum projections of two to four central planes of processed image stacks.

Measurements were done with MetaMorph 6.2r6 on the plane of an image stack that was closest to the hyphal centre. Growth speed was measured by acquisition of a DIC image followed by a time interval of 150 seconds and a DIC image stack. The cortical zone where AgExo70-GFP was enriched was approximated as a spherical cap. For measurements, the lower scaling value was set to the value of the maximal fluorescence in a 5 μm circle whose centre was located 10 μm behind the tip. Standard errors of the mean (s.e.m.) are given throughout this study.

Sample preparation for TEM was performed according to McDaniel and Roberson (McDaniel and Roberson, 2000). Mycelia were inoculated on thin dialysis membranes on AFM agar overnight at room temperature. Membranes with A. gossypii colonies were plunge-frozen in a liquid propane-ethene mixture. Freeze-substitution took place in 1% glutaraldehyde and 1% tannic acid (w/v) in anhydrous acetone at –80°C for 72 hours. After washing in acetone, the samples were warmed up stepwise in a 1% OsO₄ solution in acetone and flat-embedded in Spurr's resin (Spurr, 1969). Selected hyphae were sectioned and post-stained in 2% uranyl-acetate in 50% ethanol and in Reynolds' lead citrate (Reynolds, 1963), sections were examined using a Philips CM12S TEM (Philips Electronic Instruments, Mahwah, NJ).

M.K. thanks R.W.R. for generous support during his 4 months electron microscopy training. We are very grateful to Daniele Cavicchioli and Philipp Knechtle for providing the AgBem1-GFP strain prior to publication, to Andreas Kaufmann for sharing new A. gossypii gene targeting cassettes, to Dominc Hoepfner for the NAT1 cassette, to Sylvia Voegeli for bioinformatics support and to David Drubin for yeast strains. We thank Sandrine Grava, Andreas Kaufmann, Hans-Peter Helfer and Katrin Hungerbühler for helpful comments on the manuscript. We also acknowledge the help of Sandra Gass, Stéphanie Hauselmann and Déborah Ley in strain constructions. The work was supported by a training grant form the Swiss Federal Department of Home Affairs (ELTEM-Biotechnology) and by research grants from the University of Basel.