Characterization of AnNce102 and its role in eisosome stability and sphingolipid biosynthesis

The plasma membrane is implicated in a variety of functions, whose coordination necessitates highly dynamic organization of its constituents into domains of distinct protein and lipid composition. Eisosomes, at least partially, mediate this lateral plasma membrane compartmentalization. In this work, we show that the Nce102 homologue of Aspergillus nidulans colocalizes with eisosomes and plays a crucial role in density/number of PilA/SurG foci in the head of germlings. In addition we demonstrate that AnNce102 and PilA negatively regulate sphingolipid biosynthesis, since their deletions partially suppress the thermosensitivity of basA mutant encoding sphingolipid C4-hydroxylase and the growth defects observed upon treatment with inhibitors of sphingolipid biosynthesis, myriocin and Aureobasidin A. Moreover, we show that YpkA repression mimics genetic or pharmacological depletion of sphingolipids, conditions that induce the production of Reactive Oxygen Species (ROS), and can be partially overcome by deletion of pilA and/or annce102 at high temperatures. Consistent with these findings, pilAΔ and annce102Δ also show differential sensitivity to various oxidative agents, while AnNce102 overexpression can bypass sphingolipid depletion regarding the PilA/SurG foci number and organization, also leading to the mislocalization of PilA to septa.

The plasma membrane (PM) is involved in a variety of functions such as cell adhesion, endocytosis, exocytosis, while serving as a platform for various signalling complexes. In the last years, it has increasingly become evident that the PM is highly compartmentalized for accomplishing these roles 1,2 . Studies mainly with Saccharomyces cerevisiae and other ascomycetes focused on the lateral organization of biological membranes, have shown that PM is organized into numerous partially overlapping domains 3 , some of which are protein organized and large enough to be monitored by fluorescence microscopy. But how these domains arise is poorly understood, largely due to the technical difficulties in studying hydrophobic membranes 4 .
In the PM of yeast, three types of distinct spatial domains with different lipid and protein composition are well characterized: MCP, MCC and MCT. The first identified domain (Membrane Compartment of Pma1p, MCP) was found to contain the highly expressed plasma membrane H + -ATPase -Pma1p, which forms a network-like pattern 5 . In contrast to MCP, MCC (Membrane Compartment of Can1p, arginine permease) appears as large immobile patches of roughly 300 nm diameter forming long furrow-like invaginations that contain several transmembrane proteins 6,7 and members of the Sur7 and Nce102 families of tetraspan proteins 8,9 . MCC organization is at least in part mediated by the Nce102 protein 10 and by a cellular stable structure termed eisosome, lying underneath MCC. Each eisosome is composed of three proteins in thousands of copies, the phylogenetically related cytoplasmic Pil1 and Lsp1 and the transmembrane Sur7 protein 11 . Eisosomes in S. cerevisiae are organized/regulated by the phosphorylation of Pil1 and Lsp1 by the Pkh1/2 kinase (homologues of mammalian 3-phosphoinositide-dependent kinase) and the levels of sphingolipid Long-Chain Bases (LCBs) [12][13][14] . Sphingolipids, especially abundant complex sphingolipids, are important structural components of eukaryotic cell membranes. In addition to their structural roles, sphingolipid metabolites such as ceramides and LCBs can act as signaling molecules in many cellular processes, including cell migration, stress response, survival, apoptosis, senescence, differentiation and endocytosis 15,16 . The tetraspan protein Nce102 has been implicated as part of a sensor for sphingolipid homeostasis 10 . Membrane stress, including sphingolipid depletion, triggers eisosome proteins Slm1/2, to move out of MCC and associate with the third well characterized PM domain, MCT, Membrane Compartment of TORC2 (target of rapamycin kinase complex 2) 10,17 . The Slm1/2 proteins then recruit Ypk1 to the PM, where it is phosphorylated by Tor2 and Pkh1/2, in order to be fully activated 17,18 . Once fully activated, Ypk1 phosphorylates and consequently inactivates the endoplasmic reticulum (ER)-localized proteins, Orm1 and Orm2 17,19,20 . Furthermore, it stimulates the function of the ceramide synthase complex, by increasing the rate of the formation of ceramides and preventing hyper-accumulation of LCBs/LCBPs, thus avoiding inadvertent induction of autophagy under sufficient conditions 21 .
In the model filamentous fungus Aspergillus nidulans, PilA, PilB and SurG, homologues of the S. cerevisiae eisosome proteins Pil1/Lsp1 and Sur7, are assembled and form tightly packed structures 22 . In conidiospores and ascospores, the three proteins colocalize at the cell cortex forming stable structures that differ from the clearly distinct eisosome patches observed in S. cerevisiae 22,23 . The aim of the present study was to use the genetically tractable filamentous fungus Aspergillus nidulans for a detailed live-cell imaging and characterization of AnNce102 at different developmental stages of fungus asexual life cycle. We examined the contribution of AnNce102 to the organization/stability of eisosome foci and its possible role in sphingolipid biosynthesis and YpkA (YPK1 homologue) signaling. Our results indicate that AnNce102 colocalizes with eisosomes and affects the density/number of PilA/SurG foci in the head of germlings. Myriocin treatment, similar to annce102Δ, results in PilA foci disassembly, a phenotype that can be overcome by overexpression of AnNce102. In addition we show that the main organizers of eisosomes, the PilA and AnNce102 proteins, negatively regulate sphingolipid biosynthesis. of annce102 open reading frame with GFP and mRFP fluorophores were constructed, expressing annce102 from its endogenous promoter. Deconvoluted Z stacks of AnNce102 in quiescent conidia showed a plasma membrane staining pattern that colocalizes with PilA and partially localizes in the remainder of the membrane (Fig. 1A). AnNce102 localizes additionally in intracellular structures resembling the Endoplasmatic Reticulum (ER), as shown by its colocalization with the ER chaperone, ShrA 24 (Fig. 1B). During isotropic growth, the AnNce102 ER fluorescence signal diminishes and it disappears before the emergence of the germination tube ( Fig. 1C and below). In germlings, AnNce102 foci are mostly confined to eisosomes of the hyphal head and to vacuoles, as stainable with the vacuolar tracer, CMAC. Additionally, AnNce102 is clearly detected in septa of germlings (Fig. 1C, white arrow) and colocalizes with some PilA foci along the hyphal plasma membrane (Fig. 1C, black arrows). This temporal and spatial AnNce102 distribution resembles that of the SurG tetraspan eisosomal protein (Fig. 1D,E) 22 . Moreover using time-lapse live-cell imaging, we show that AnNce102 forms stable structures with low mobility at fungal plasma membrane (Fig. 2).

Nce102 of
Deletion of pilA leads to low levels of AnNce102 and SurG in growing germlings. In yeast cells, deletion of PIL1 disrupts MCCs and all remaining eisosome proteins investigated so far coalesce into one or a few punctate foci at the cell periphery, referred to as eisosome remnants 7,13 . In addition, Sur7 and Nce102 localize to a few bright clusters and a pool diffusely distributed over the plasma membrane 10,13 . On the contrary, deletion of pilA in quiescent conidia of A. nidulans has a minor effect on AnNce102 and SurG distribution. Both proteins appear to be localized less on the membrane and more in vacuoles compared to wild type cells (Fig. 3A,D). Similarly, almost no fluorescence is detected on the membrane during isotropic growth (Fig. 3E), while in germlings both proteins are localized exclusively in vacuoles and endosomes (Fig. 2F). RT-PCRs in total RNA extracted from quiescent conidia (0 h) and germlings (14 h) of a wild type strain and pilAΔ revealed that mRNAs of both annce102 and surG are abundant throughout the time-interval tested without significant differences (Fig. 3G). In contrast, western blot, using protein extracts from quiescent conidia (0 h) and germlings (14 h) confirmed microscopic observations. More specifically, deletion of pilA causes enhanced degradation of AnNce102-GFP and SurG-GFP in the vacuoles, as evidenced by the diminished signal of intact GFP-tagged proteins and the increased quantity of free GFP, even in quiescent conidia (Fig. 3H). We should also mention that deletion of pilB did not affect the membrane localization of AnNce102 and SurG of young germlings, while pilAΔ pilBΔ double mutants did not show any difference from pilAΔ single mutants ( Supplementary Fig. S1). Together, these data suggest that normal eisosomal organization requires PilA.
Deletion of annce102 leads to fewer PilA and SurG foci in the head of germlings. annce102 deletion mutants displayed normal growth at 25, 30, 37, and 42 °C (see Materials and Methods), while conidia exhibited properties of swelling and polarity establishment (time of germination tube appearance) indistinguishable from that of a wild type strain at all temperatures tested ( Supplementary Fig. S2). Contrary to S. cerevisiae, where nce102Δ mutants displayed altered localization of Sur7-GFP, Pil1-GFP, Can1-GFP, and ergosterol 8,10 , deletion of the annce102 gene has no significant effect on PilA-mRFP and SurG-GFP distribution in quiescent conidia of A. nidulans, with both proteins colocalizing on the membrane (Fig. 4A,B). In annce102Δ germlings however, PilA and SurG displayed an uneven distribution compared to wild type (Fig. 4C,D). Using the spot tool of the Imaris software, we found that PilA forms 3-fold fewer foci in the head of annce102Δ germlings (Fig. 4E). Interestingly, a similar effect is observed in the presence of sub-lethal doses of myriocin ( Fig. 5A) (inhibitor of serine palmitoyltransferase, the enzyme catalyzing the first step in sphingolipid biosynthesis), where in addition no colocalization of PilA-mRFP and SurG-GFP foci is observed (Fig. 5Bi). The similarity of the phenotypes suggests that, like in yeast 10 , myriocin could affect the organization of eisosomes through AnNce102. To add to the significance of the above, the effects of myriocin addition and annce102 deletion on the number of PilA foci were not additive (Fig. 5C), while overexpression of annce102 can protect from myriocin-induced eisosome disassembly (see below).
The myriocin effect on eisosome localization may be due to the lack of LCBs, complex sphingolipids or a more general response to sphingolipids overall. To distinguish between these possibilities, we tested the effect of Aureobasidin A (AbA), a specific inhibitor of complex sphingolipids 25 (Fig. 5A). In PilA-mRFP/SurG-GFP cells treated with AbA, although we observed less pronounced reduction of PilA foci in the head of conidia-derived germlings, the treated cells displayed enhanced degradation of SurG ( Fig. 5 Bii and Biii, compare with 4E), suggesting that proper eisosomal organization depends on the overall level of sphingolipids. Together, the above data show that normal eisosomal organization requires AnNce102 and suggests that this protein is involved in sphingolipid biosynthesis.
Having shown that the AnNce102 protein affects eisosomal foci number in the head of germlings, and given that PilA is the only eisosomal protein displaying a punctate distribution pattern along the hypha 22 , we attempted to identify other proteins that could potentially affect the distribution/number of PilA foci along hypha. We chose to study proteins that exhibit high identity to the amino acid sequence of AnNce102 MARVEL-containing protein. In silico analysis of the genome of A. nidulans revealed 2 sequences, AN8278 and AN8422, designated mrvA and mrvB, showing 26.9% and 23.9% identity to AnNce102 and displaying 9.0e-07 and 6.0e-03 E values to AnNce102, respectively. However, strains with deletions in one or both genes exhibited normal growth in all conditions tested and no changes in the topology of PilA in both conidia and germlings were observed ( Supplementary Fig. S3). These results suggest that mrvA and mrvB do not participate in eisosome stability/organization.

Overexpression of AnNce102 restores PilA foci number in the presence of myriocin and leads to ectopic localization of PilA foci into septa.
In order to test the effect of annce102 overexpression on PilA foci number in myriocin-treated cells, a conditional annce102 mutant strain was constructed, alcA::annce102::mRFP, expressing annce102 under the control of the alcohol dehydrogenase (alcA) promoter, which is repressed by glucose and induced by ethanol 26 . Under repressing conditions the number of PilA foci in the head of germlings is similar to that of annce102Δ cells (Fig. 6Ai). Under inducing conditions, AnΝ ce102 is diffusely distributed to the periphery of cells, although some colocalization with PilA foci can be detected in the head of germlings. Most importantly, overexpression of AnNce102 in the presence of myriocin results in normal PilA foci number in the head of germlings (Fig. 6B).
Interestingly, under conditions of induced expression of AnNce102, PilA foci are also clearly localized to septa (Fig. 6, white arrows). As shown in Fig. 6C (white arrows), PilA foci do not localize to calcofluor white-stained septa of A. nidulans. However, overexpression of AnNce102 causes mislocalization of PilA foci to septa without however, affecting their morphology. These data confirm the involvement of AnNce102 in sphingolipid sensing and demonstrate the crucial role of AnNce102 regarding PilA foci number and topology.

Sphingolipid depletion and ypkA repression increase ROS levels. In S. cerevisiae it has been
shown that perturbation of sphingolipid metabolism leads to loss of cell viability and Reactive Oxygen Species (ROS) accumulation 16 . In addition, deletion of nce102 confers higher sensitivity to oxidative stress 27 . Thus, we have investigated whether sphingolipid depletion regulates ROS accumulation in A. nidulans as well. Toward this end, we studied cells treated with myriocin and Aureobasidin A, specific inhibitors of sphingolipid biosynthesis 25,28 , as well as cells genetically blocked for sphingolipid biosynthesis. 20% of myriocin and 36% of AbA treated cells displayed increased ROS levels, visualized by the ROS indicator dye 2′ , 7′ -dichlorofluorescin diacetate (DCF) (Fig. 7A). In addition, 10% of germlings from basA1 cells possessed elevated ROS levels at 42 °C (Fig. 7C). These cells carry a point missense mutation in basA, the A. nidulans SUR2 homologue of S. cerevisiae, which results in thermosensitivity of C4-hydroxylase, the enzyme catalyzing the conversion of dihydrosphingosine to phytosphingosine 29 . This increase in ROS levels was relieved by incubated basA1 cells in the presence of the ROS scavenger N-acetyl cysteine (NAC) (Fig. 7B). Ιt is worth mentioning that in untreated wild type cells grown under these conditions, the percentage of ROS was extremely low. Consistent with the idea that sphingolipid depletion causes ROS accumulation, addition of phytosphingosine (PHS), a 4-hydroxylated LCB, only reduced ROS production in myriocin and not AbA treated cells. Notably, PHS administration in AbA-treated cells resulted in augmented ROS levels, suggesting that not only depletion of complex sphingolipids but also accumulation of LCBs can result to further production of ROS 30 (Fig. 7B). Sphingolipid biosynthesis in yeast is regulated by the Ypk1 kinase 18,19,21,[31][32][33] . Interestingly, repression of YPK1 homologue in A. nidulans, using the regulatable promoter niiA (niiA::ypkA strain kindly provided by Dr. Goldman 34 ) by ammonium resulted in a significant increase in ROS levels, evident after 12 h of growth in presence of ammonium as sole nitrogen source at 25 °C (Fig. 7D).
To further investigate the interrelationship of ypkA with sphingolipid biosynthesis and ROS production, double niiA::ypkA basA1 mutant strains were constructed. These double mutants displayed a synthetic lethal phenotype when grown under repressing conditions at the non-permissive temperature (42 °C), although under inducing conditions at 42 °C niiA::ypkA partially suppressed basA1 lethality The number of PilA foci in annce102 + and annce102Δ cells was counted using the spot tool of the Imaris software and is shown below the images, as a graph with averages presented as mean ± SEM, where n represents the number of cells examined from at least three experiments. Scale Bars 3 μ m.
Having established the conditions and tools, we sought to examine whether AnNce102 and PilA are directly involved in oxidative damage. In our analysis we also compared the use of two different selection markers to create the deletion of annce102 in order to exclude possible marker effects (see Materials and Methods). In all cases, mutants lacking annce102 showed similar growth, regardless of selection marker used. Cell viability of A. nidulans mutants growing in the presence of H 2 O 2 , menadione or paraquat was assayed (Fig. 8). Both drugs, menadione and paraquat are cycling agents that generate superoxide ions 35 . Interestingly, cells with deletion of annce102 were more sensitive to H 2 O 2 stress and pilAΔ and/ or annce102Δ mutants were more resistant in the presence of menadione (Fig. 8). On the other hand, all mutant strains were found to be hypersensitive to paraquat (Fig. 8).  29 . Genes and inhibitors used in this study are shown in red boxes. (B) Deconvoluted confocal images of PilA-mRFP and SurG-GFP, shown as equatorial sections and volumetric view of confocal stacks, treated for 5 h with 20 μ g/ml myriocin (i) and 0,25 μ g/ml Aureobasidin A (AbA) (ii). The number of PilA foci was counted using the spot tool of the Imaris software and is shown as a graph with averages presented as mean ± SEM, where n represents the number of cells examined from at least three independent experiments (iii) (C) Representative deconvoluted confocal images of wild type and annce102Δ cells expressing PilA-GFP, treated for 5 h with 20 μ g/ml myriocin. The number of PilA foci was counted as above and is shown as a graph with averages presented as mean ± SEM, where n represents the number of cells examined from at least three independent experiments. Scale Bars 5 μ m.
Scientific RepoRts | 5:15200 | DOi: 10.1038/srep15200 The above data show that ANnce102 or/and PilA play a role in protection against oxidative stress generated by paraquat and H 2 O 2 , also suggesting a role of the YpkA kinase in sphingolipid biosynthesis pathway and confirming the important role of YpkA and sphingolipids in redox homeostasis.
AnNce102 and PilA genetically interact with YpkA and BasA, key players in sphingolipid metabolism. In order to test whether deletion of PilA and AnNce102 affects ROS levels through regulation of YpkA activity, we created double and triple mutants of pilAΔ and/or annce102Δ with niiA::ypkA. Simultaneous deletion of both pilA and annce102 results in impaired growth of a niiA::ypkA in ammonium at 25 °C, and, most importantly, seems to partially suppress the defects of ypkA downregulation at 42 °C (Fig. 9A). The growth of all mutants is well correlated with their intracellular ROS levels detected by DCF at 25 °C (Fig. 9B), where niiA::ypkA annce102Δ pilAΔ triple mutants display 3 fold more ROS at 25 °C. At 42 °C no statistically significant differences in ROS levels were detected. In order to test whether the above differences in growth and ROS accumulation are caused by the regulation of sphingolipid biosynthesis, we introduced the pilAΔ and/or annce102Δ mutations in a basA1 strain. As shown in Fig. 9C, basA1 mutants display slightly impaired growth at the semi-permissive temperature (37 °C), which is furthermore enhanced at non-permissive temperature (42 °C). Most remarkably, the double and triple mutants grow significantly better than the single basA1. Consistent with the above, annce102Δ and/or pilAΔ were found to be partially resistant to myriocin and AbA treatment at high temperatures (Fig. 9D). Curiously, addition of the ROS scavenger N-acetyl cysteine (NAC) effectively concealed any growth differences among basA1, niiA::ypkA, annce102 and pilA mutants observed above (Figs 9A,C), although no statistically significant differences in ROS levels were detected among mutants. Moreover, we quantitatively determine the inhibitory effects of the inhibitors of sphingolipid metabolism, myriocin and AbA, in WT, pilAΔ, annce102Δ , pilAΔ annce102Δ and alcA::annce102 cells at 37 °C, in the presence of MM+ 2% glucose or MM+ 2% glycerol+ ethanol. Myriocin was the least active against A. nidulans with a minimal inhibitory concentration (MIC) of 80 μ g/ml in the presence of glucose and 60 μ g/ml in the presence of glycerol+ ethanol. The MIC of AbA was 4 μ g/ml in the presence of glucose Scientific RepoRts | 5:15200 | DOi: 10.1038/srep15200 and 2 μ g/ml in the presence of glycerol+ ethanol. In these conditions we were not able to detect significant differences in the resistance or sensitivity of mutants compared to wild-type ( Supplementary  Fig. S4).

Discussion
Several studies so far have documented the role of Nce102 proteins and despite their conservation among the ascomycetes 36,37 , the functional role of these proteins may diverge. For instance, in Ashbya gossypii, the orthologue of Nce102 despite colocalizing with eisosomes is not needed for eisosome stability and polar growth 38 . In Aspergillus fumigatus AfuNce102 deletion mutants showed a clear delay in conidiophore formation and severely affected sporulation, without affecting the virulence of the fungus 37 . Deletion of Nce102 from Candida albicans caused a 2-fold decrease in MCC/eisosomes. However, the  cells displayed normal punctate localization of Lsp1-GFP and Sur7-GFP and showed decreased virulence by forming abnormal hyphae in mice 36 .
In this study, we present evidence that PilA and Nce102 are the main eisosomal organizers of A. nidulans. In particular, we showed that AnNce102 colocalizes with PilA at the PM and its temporal and spatial distribution resembles that of the SurG eisosomal protein. In addition, we observed that deletion of pilA has enormous impact on the stability of AnNce102 and SurG in germlings, where both proteins localized exclusively in vacuoles and endosomes. Despite the fact that increased endocytosis and enhanced direct vacuolar sorting of SurG and AnNce102 could both rationalize the data presented herein, we favor the hypothesis that the absence of PilA, which is mostly found at the plasma membrane, most probably destabilizes the otherwise stable foci of AnNce102 (and SurG) rather than it affects their intracellular trafficking. This is consistent with the proposed role of eisosomes in protecting Can1 from endocytosis in yeast 8 , which is however currently under debate 6 .
On the other hand, deletion of AnNce102 affects the older PilA/SurG eisosomal pool, resulting in fewer foci only in the head of germlings and without affecting eisosomes along hyphae. This phenotype resembles myriocin treatment and, similarly to S. cerevisiae 10 , can be overcome by overexpression of AnNce102. The above results suggest that two different populations of PilA foci exist in A. nidulans germlings; one at the head of the germlings that requires AnNce102 for proper organization and is sensitive to sphingolipid levels and another in the hyphae that does not respond to the above and possibly has different function and composition. Interestingly, deletion of annce102 or pilA affects each other topology/foci number mostly in growing germlings. This temporal regulation could be related to the potential role(s) of eisosomes in sphingolipid and/or PI(4,5)P 2 regulation 1-4 and present study , recruitment of signaling molecules 5 and the greater need of membrane biogenesis and compartmentalization in actively growing and not quiescent cells. Furthermore, overexpression of AnNce102 in A. nidulans led to ectopic localization of PilA foci into septa, a result that further confirms the crucial role of AnNce102 concerning eisosome organization, also suggesting a functional role of eisosomes related to septation. Consistent with this, deletion of sur7 in C. albicans resulted in ectopic localization of the cdc12 septin from the bud neck 39 .
In S. cerevisiae Pil1 down-regulates the Ypk1 pathway 14 and Nce102 has been implicated as a sensor for sphingolipid abundance and a negative regulator of pkh1 activity 10 . Our data support a subtle role for AnNce102 and PilA in the regulation of sphingolipid biosynthesis in A. nidulans. Indeed, we herein show that deletion of either pilA or annce102 can partially suppress genetic (basA1 and niiA::ypkA) or pharmacological (myriocin and AbA) inhibition of sphingolipid biosynthesis at high temperatures (Fig. 9). Consistent with this, it has been shown previously that heat shock leads to a transient increase in sphingolipid production and that sphingolipid production is important for heat shock survival 40,41 . The fact that MIC measurements did not display significant differences, suggests a minor role of PilA and AnNce102 proteins in the regulation of sphingolipid metabolism (Fig. S4). An intriguing alternative explanation is that the difference observed in agar plates, in which A. nidulans colonies differentiate by asexual sporulation, cannot be observed by MIC measurements, which have been done in liquid media only allowing the growth of mycelia. Consistent with the above, the expression of basA was shown to increase upon begin of sexual and asexual development and the basA1, barA and lagA mutants have been shown to negatively affect the asexual/sexual ratio of sporulation in A. nidulans 29 , strongly indicating that asexual sporulation requires increased levels of sphingolipid biosynthesis. In light of the above, our results presented in Fig. 9 can be rationalized by developmental stage-specific genetic interaction of eisosomes and the sphingolipid biosynthesis pathway. In support of this, deletion of ypkA is known to result in very slow-growing, non-sporulating cells, while the defective growth of a niiA::ypkA strain in repressing conditions can be overcome by PHS supply 34 suggesting that YpkA, possible through the regulation of sphingolipid metabolism, indeed participates in asexual sporulation.
Another important aspect related to Nce102 is its implication with oxidative stress 27 . Reactive oxygen species (ROS) are versatile molecular species and radicals that are poised at the core of a sophisticated network of signaling pathways and act as core regulators of cell physiology and cellular responses to environment 42 . ROS also help to monitor/modulate cellular processes that range from different cell fates 43 to apoptosis 44 , from regulation of Apical Dominance 45 to actin polarization 46 , suggesting that a kind of 'ROS rheostat' exists in cells. It is also known that oxidants can influence regulatory proteins to modulate sphingolipid metabolism and also appear to serve as upstream/downstream messengers for sphingolipid signaling [47][48][49] . In this study, we observed that, deletion of AnNce102 results in hypersensitivity to H 2 O 2 . However, treatment of pilAΔ and/or annce102Δ strains with 2 superoxide generators, had opposing effects on growth, where pilAΔ , annce102Δ and pilAΔ annce102Δ strains were more resistant in the presence of menadione and less resistant to paraquat. These results suggest a mechanism other than superoxide formation being responsible for this discrepancy. Moreover, we demonstrated that pharmacological (using myriocin and AbA) or genetic (using basA1 mutant) inhibition of sphingolipid synthesis resulted in a significant increase in ROS levels. Interestingly, repression of YpkA (YPK1 homologue) expression resulted also in an increase in ROS, evident after 12 h of repression. The kinetics of this increases correlate well with the decreased radial growth, the delayed conidial germination, a deficiency in polar axis establishment and the intense branching observed after the germination of conidia of the niiA::ypkA strain grown under repressing conditions 34 . Furthermore, we showed that niiA::ypkA basA1 double mutant strain displayed a synthetic lethal phenotype when grown under repressing conditions at the non-permissive temperature (42 °C), similar to synthetic lethality phenotype of barA1 niiA::ypkA mutants grown in the presence of ammonium at 37 °C 34 . In addition double mutants show higher ROS levels than single mutants at the same conditions, confirming a previous suggested role of YpkA kinase in sphingolipid biosynthesis pathway 34 and suggesting an important role of YpkA and sphingolipids in redox homeostasis. Concerning the role of eisosomes/AnNce102 in YpkA/sphingolipid signaling we showed that pilAΔ and/or annce102Δ strains were partially resistant to pharmacological or genetic reduction of sphingolipid synthesis at high temperatures. Under these conditions, addition of the ROS scavenger N-acetyl cysteine (NAC) effectively concealed any growth differences among mutants, although no statistically significant differences in ROS levels were detected (Supplementary Fig. S5). A possible explanation could be that the ROS produced in these conditions cannot be detected by DCF, used in the present study. Interestingly we found niiA::ypkA annce102Δ pilAΔ triple mutant to be more sensitive at 25 °C under repressing conditions, and the only strain that displayed 3 fold more ROS than single or double mutants. Within this context, nce102Δ mutants of C. albicans were defective in forming hyphae and invading low concentrations of agar, likely due to a defect in actin organization, whereas Ypk1 has been proposed to regulate actin polarization by controlling ROS accumulation 46 . Since repression of YpkA mainly causes deficient polar axis establishment and intense branching 34 , an intriguing explanation could be that besides the role of PilA/AnNce102 in sphingolipid signaling, these proteins are likely to be involved in the regulation of cytoskeleton polarization through YpkA and a mechanism that involves ROS.
Previously we have shown that pilAΔ mutants, pilAΔ pilBΔ double mutants and surGΔ mutants displayed resistance to inhibitor of ergosterol synthesis, itraconazole 22 . In addition, it was recently shown that exogenous addition and/or improper accumulation of LCBs resulted in altered susceptibility to inhibitors of β -1,3-glucan synthesis 50 , echinocandins, in various fungi 51 . All these data support the possibility that eisosomal proteins modulate the mode of action of these groups of antifungal drugs by regulating sphingolipid biosynthesis rather than directly regulating β -1,3-glucan synthase as was hypothesized by Edlind and Katiyar 52 , making these proteins ideal candidates for studying sphingolipid signaling, antifungal resistance/susceptibility and their underlying mechanisms.

Methods
The entire annce102, mrvA, and mrvB open reading frames (ORFs) (627 bp [AN7683], 570 bp [AN8278], and 747bp [AN8422], respectively) were replaced in appropriate strains (see Supplementary Table S1) by the A. fumigatus, riboB (Afribo), pyrG (AfpyrG) or pyroA (Afpyro) genes, using the fusion PCR gene replacement method 53 and standard molecular cloning techniques. Recipient strains also carried a nkuAΔ mutation that results in a dramatically decreased frequency of heterologous integration events into the A. nidulans genome 54 . Cassettes containing the annce102::sgfp and annce102::mrfp fusions, expressed under the control of the endogenous promoter or under the alcA (alcohol dehydrogenenase gene) promoter, were constructed by joining three different PCR fragments, as previously described 53 . For the AnNce102-GFP and AnNce102-mRFP fusions, the GFP and mRFP were separated from the ORFs by a quintuple Gly-Ala di-peptide repeat (5GA linker). Cassettes for gene replacement were constructed and amplified using Kapa-HiFi (Kapa Biosystems) DNA polymerase and the primers of Supplementary  Table S2. In-locus integration events in all transformants used in this study were confirmed by Southern analysis and PCR (see below).
RNA samples were prepared as previously descripted 23 RNA was isolated from quiescent conidia (0 h) and germlings (14 h), using the TRI Reagent (GIBCO-BRL) kit according to the instructions of the manufacturer. RNA samples were further purified according to the RNeasy Mini Protocol for RNA Cleanup, in the RNeasy Mini Kit (Qiagen). To avoid contamination with genomic DNA, 10 μ g of each RNA sample were treated and cleaned up with TURBO DNA-freeTM kit (Ambion). The absence of DNA contamination was verified with conventional PCRs using specific-, surG, annce102 and 18S rRNA set of primers and at least 3 μ g of each RNA sample as template, which were amplified for 40 cycles. Primers were designed to cross at least an intron and expected to produce amplicons of 355 bp (surG cDNA), 478 bp (annce102 cDNA), and 280 bp (18S ribosomal RNA gene). The quality of RNA was confirmed in a conventional 2% w/v agarose gel stained with ethidium bromide (Et-Br) (10 μ g/ml). The concentration of each RNA sample was calculated using the nanodrop apparatus (ND-1000 Spectrophotometer) according to the instructions of the manufacturer. Approximately 3 μ g of each RNA sample were used for reverse transcription using the SuperScriptTM II RNase H-reverse transcriptase (Invitrogen), according to the instructions of the manufacturer. For semi-quantitative analysis of transcript levels, cDNA was diluted 10-, 100-, 1000-fold. The number of cycles required to produce detectable difference in band intensity between dilutions (to avoid saturation) was determined with Et-Br staining.
For most figures, representative images of the equatorial sections and/or volumetric views of confocal stacks are shown. Images were acquired sequentially as Z-series (step size 0.2 μ m) and processed/analyzed using the Fiji/imagej software (http://imagej.nih.gov/ij/download.html) or the Imaris 7.2.4 (Bitplane AG, Zurich, Switzerland). For colocalization assessment z-stacks were deconvolved using a three-dimensional (3D) blind algorithm [Autoquant X (Media Cybernetics, Silver Spring, MD)]. Colocalization was measured using Imaris software package (Bitplane, Zurich, Switzerland). The threshold of each channel used to quantify colocalization was determined automatically, according to the method of Costes 56 . Colocalization was defined as the overlap of two channels in three dimensions and was calculated by the program automatically. Colocalization was analyzed by using Manders colocalization coefficients (M1 and M2) that describe the colocalization of molecules with respect to an individual channel, and increase from 0 to 1 with rising colocalization and visual inspection of the spatial relations (intensity spatial profile). 2D histograms (plotted pixels of a two-channel image) were obtained using Imaris (Bitplane). Colocalization was represented as new channel (colocalization channel) corresponding to the colocalized voxels. PilA foci in germlings were counted using spot detection tool of the Imaris (Bitlane) software. A minimum diameter of 0.3 μ m and an automatic threshold detection of the software were used, adjusted manually to identify all visible PilA foci.
Time-lapse experiments were conducted at 26 °C. Cells were imaged by the use of Leica TCS SP5 (Leica Microsystems Ltd., Milton Keynes, UK) confocal microscope with a Leica HCX PL APO CS 63x/1.4 NA oil immersion lens. The images were analyzed with the Imaris software package (Bitplane, Zurich, Switzerland).
Total protein extracts were prepared as previously described 23,57 , with the following modifications. Approximately 500 mg of liquid N-grinded conidiospores or young mycelia (14 h, 25 °C) were resuspended in 1.5 ml ice-cold precipitation buffer (50 mM Tris-HCl pH 7, 50 mM NaCl, 1 mM EDTA), supplemented with a protease inhibitor cocktail (Sigma) and 1 mM phenylmethylsulfonyl fluoride (PMSF). TCA was added (1/8 of total V), the suspension was briefly vortexed, followed by centrifugation (15000 g, 10 min, 4 °C). The pellet was washed twice with ice cold acetone, heat-dried and dissolved in 0.5 ml extraction buffer (150 mM Tris-HCl pH 8, 50 mM NaCl, 1% w/v SDS, 1mM EDTA and protease inhibitor cocktail). Protein concentrations were determined by the method of Bradford. In each case 30-50 μ g protein were fractionated on a 10% (w/v) SDS-polyacrylamide gel and electroblotted (Mini Protean Tetra cell, Bio-Rad) onto a polyvinylidene difluoride (PVDF) membrane (Macherey-Nagel, Lab Supplies Scientific SA, Hellas) for immunodetection. The membrane was treated with 2% (w/v) non-fat dry milk, and immunodetection was performed using a primary mouse anti-GFP monoclonal antibody (Roche), a mouse anti-actin monoclonal (C4) antibody (MP Biomedicals Europe, Lab Supplies Scientific SA, Hellas) and a secondary rabbit anti-mouse IgG horseradish peroxidase (HRP)-linked antibody (Cell Signaling). Blots were developed by the chemiluminescent method using the LumiSensor Chemiluminescent HRP Substrate kit (GenScript USA Inc, Lab Supplies Scientific SA), an enhanced chemiluminescence reagent (Amersham Bioscience) and SuperRX Fuji medical X-Ray films (FujiFILM Europe, Lab Supplies Scientific SA, Hellas).
Myriocin and Aureobasidin A susceptibility assays. The susceptibility assays were performed by measuring the MIC of each molecule in 96-well flat bottom plates, using a slightly modified protocol previously described 58 . Briefly, the assay mixture was prepared by adding 1 volume of conidial suspension (10 5 conidia ml −1 in 0.05% Tween 20) to 12 volumes of assay medium (MM+ 2% glucose or MM+ 2% glyc-erol+ ethanol). Two-fold dilutions of each drug were prepared with this assay mixture. The plates were incubated at 37 °C for 24 h. After incubation, the culture medium was removed from each well and the biofilms were washed three times with Milli-Q water. 130 ml of 0.01% (w/v) crystal violet solution (Sigma Aldrich) was added to each well for 20 min at room temperature. The solution was then removed and the biofilms were washed until the supernatant was clear. The plates were air-dried and the absorbance at 560 nm was measured in a Tecan Infinite M-1000 PRO plate reader (Tecan Systems Inc., San Jose, CA).