Lsp1 partially substitutes for Pil1 function in eisosome assembly under stress conditions

The eisosome is a plasma membrane-associated protein complex described in numerous fungal species, whose core consists of Pil1 family proteins (Walther et al., 2006). Membrane association of Pil1-like proteins is mediated by their Bin/Amphiphysin/Rvs (BAR) domain and induces negative membrane curvature. Moreover, these proteins self-assemble when associated with membranes, leading to liposome tubulation in vitro and the formation of the hemitubular core of the eisosome in vivo (Ziółkowska et al., 2011; Karotki et al., 2011; Olivera-Couto et al., 2011; Suarez et al., 2014). Therefore, plasma membrane microdomains associated with the eisosome, known as the membrane compartments of Can1 (MCC), have a typical furrow shape (Stradalova et al., 2009).

The composition of the MCC and eisosome (hereafter MCC/eisosome) is highly variable. It changes depending on instantaneous conditions such as nutrient availability, membrane potential and membrane tension (Malínská et al., 2003; Grossmann et al., 2007; Berchtold et al., 2012; Gournas et al., 2016; Riggi et al., 2018; Malinsky and Opekarová, 2016; Zahumensky and Malinsky, 2019). The dynamic adaptability of the MCC/eisosome structure suggests that it has a role in cellular stress perception and response, and indeed, numerous stress-related phenotypes have been reported in cells defective in MCC/eisosome formation to date (Young et al., 2002; Zhang et al., 2004; Fröhlich et al., 2009; Dupont et al., 2010; Mascaraque et al., 2013). Fungi in nature are constantly facing different kinds of stress. It is, therefore, not surprising that not only the Pil1 family proteins but also other components of the MCC/eisosome have been conserved during phylogenetic evolution (Stradalova et al., 2009; Loibl et al., 2010; Moreira et al., 2012; Grousl et al., 2015; Lee et al., 2015).

Cells of many fungal species contain two homologous eisosome core proteins with slightly different biological functions. In the filamentous fungus Aspergillus nidulans, the two eisosomal organizers PilA and PilB form eisosomes at the spore periphery, whereas mycelial eisosomes only contain PilA (Vangelatos et al., 2010). In the fission yeast Schizosaccharomyces pombe, eisosomes in the spore membrane are composed exclusively of Pil2, whereas eisosomes at the plasma membrane of cells generated during the mitotic cell cycle, including the membrane of the ascus, contain only Pil1 (Kabeche et al., 2011). In the budding yeast Saccharomyces cerevisiae, distinct phosphorylation patterns of two Pil1 family proteins in the presence of sphingolipid precursors, long-chain bases (LCBs), formed the basis for the naming of these proteins: Pil1 – phosphorylation inhibited by LCBs, and Lsp1 – LCBs stimulate phosphorylation (Zhang et al., 2004). Under normal laboratory conditions (cultivation at 30°C in a nutrient-rich medium), Pil1 is dominant and sufficient to produce the normal eisosomal pattern in S. cerevisiae, which consists of several tens of evenly distributed and randomly oriented 250–300 nm long eisosomes per cell. Although no structural phenotype is observed in lsp1Δ cells, deletion of PIL1 has a pronounced effect on the plasma membrane structure. Instead of regular eisosomes, rare (usually 1–2 per cell) collapsed structures containing MCC/eisosome-specific proteins including Lsp1 (eisosome remnants) that are up to 1 µm in size are formed in the plasma membrane of pil1Δ cells (Walther et al., 2006; Stradalova et al., 2009). The role of Lsp1 becomes apparent as the cell culture ages and consumes available nutrients. Under these conditions, eisosomes become larger and more numerous, which requires the presence of Lsp1 (Gournas et al., 2018).

The long all-trans acyl chains contained in sphingolipid molecules increase plasma membrane order (Boggs, 1987; Simons and Ikonen, 1997; Aresta-Branco et al., 2011; Vecer et al., 2014; Herman et al., 2015). Therefore, the cell increases their abundance in the membrane in situations that challenge the membrane integrity, such as increased temperature or membrane tension. Eisosome assembly, which requires proteins of the Pil1 family, has also been shown to dynamically respond to these challenges (Berchtold et al., 2012). The direct involvement of Pil1-like proteins in the regulation of stress adaptation has been repeatedly reaffirmed in recent studies of cells lacking any of the Pil1-like proteins (Yang and Tavazoie, 2020; Sakata et al., 2022).

A specific role of Pil1 phosphorylation in the eisosome assembly has been confirmed (Walther et al., 2007; Luo et al., 2008). Minor participation of Lsp1 in eisosome formation in S. cerevisiae has been explained in terms of its reduced ability to bind the plasma membrane (Walther et al., 2006; Olivera-Couto et al., 2011). Given the aforementioned difference in LCB-dependent phosphorylation of Pil1 and Lsp1 proteins, we tested the hypothesis that Lsp1 can substitute for Pil1 function in eisosome assembly under stress conditions, especially when sphingolipid abundance in the plasma membrane is affected. We find that Lsp1 is able to form eisosomes independently of Pil1 and does so under conditions that decrease the activity of the serine palmitoyltransferase (SPT), such as upon its direct inhibition or hyperosmotic shock. We document that Lsp1 requires the presence of either the tetraspan protein Nce102 accumulated in the MCC (Grossmann et al., 2008; Vaskovicova et al., 2020) or the soluble eisosome stabilizer Seg1 (Moreira et al., 2012; Vaskovicova et al., 2015) to do so.

In the absence of Pil1, Lsp1 and other MCC/eisosome components accumulate in eisosome remnants in cells cultivated under normal conditions (Walther et al., 2006; Stradalova et al., 2009). Using fluorescently labeled Lsp1 as the compartment marker, we monitored the morphology of eisosome remnants in pil1Δ cells under conditions of membrane stress, induced either by shifting the exponentially growing yeast cultures from optimal 28°C to 37°C for 2 h or by the addition of 1 M NaCl to the growth medium. Both heat and salt stress conditions led to remarkable changes in the cellular distribution of Lsp1–GFP. Specifically, eisosome remnants disintegrated and multiple smaller-sized foci of high fluorescence intensity formed instead (Fig. 1A,B). These well-defined Lsp1–GFP foci were distributed almost evenly along the plasma membrane, had comparable brightness to each other and were of the same size and shape (Fig. S1A–D). Their number was lower compared to the wild-type eisosomes, but significantly higher than the number of eisosome remnants under both tested stress conditions (Fig. 1A,B). In addition, quantification of the plasma membrane/cytoplasm fluorescence intensity ratio and the mean cellular GFP intensity showed a clear increase in both parameters following cultivation in the presence of 1 M NaCl (Fig. 1C,D). These results indicate an increased association of Lsp1–GFP with the plasma membrane and an elevated level of the protein in response to salt stress, respectively. We confirmed the latter by using western blot analysis (Fig. 1E).

To determine whether the Lsp1-containing foci formed in response to heat and salt stress are eisosomes, we checked one of the main distinguishing features of these structures – their ability to bend the plasma membrane into the characteristic furrow shape. On freeze-etched replicas, we compared the plasma membrane morphology in pil1Δ cells under normal and stress conditions. We found that whereas only irregular, convoluted membrane-containing eisosome remnants in otherwise flat plasma membrane could be detected under normal conditions, salt stress induced the formation of typical wild-type-like linear furrow invaginations in the plasma membrane of pil1Δ cells (Fig. 2). Heat stress exposure likewise induced eisosome formation in the pil1Δ cells (data not shown). We, therefore, conclude that the Lsp1–GFP foci described above, formed in response to heat and salt stress, correspond to eisosomes.

Given that we have found an increase in LSP1 expression in response to salt stress (Fig. 1E), we tested whether elevating the Lsp1 level is sufficient to induce the formation of Lsp1 eisosomes in pil1Δ cells even in the absence of stress. We increased LSP1 expression by exchanging its native promoter for the tetO₇ promoter using the pCM225 cassette, resulting in a threefold increase in the mean cellular Lsp1–GFP fluorescence (Fig. S1E,G). The overexpression of LSP1 led to increased accumulation of the protein within the eisosome remnants but did not trigger eisosome formation (Fig. S1E,F). When exposed to heat or salt stress, the pil1Δ cells overexpressing LSP1 formed eisosomes, analogous to pil1Δ cells expressing LSP1 from its native promoter (compare Fig. S1E,F with Fig. 1A,B). We conclude that the principle underlying the disintegration of eisosome remnants and formation of well-defined Lsp1 eisosomes in response to stress is not a change in Lsp1 protein abundance, but some other aspect of the stress response.

It has been reported that both heat and salt stress induce changes in the sphingolipid composition of the yeast plasma membrane (Dickson et al., 1997; Jenkins et al., 1997; Manzanares-Estreder et al., 2017; Zahumenský et al., 2022). We, therefore, investigated whether specific modifications of the cellular sphingolipid content could induce the dissolution of the eisosome remnants in the pil1Δ strain and the formation of the Lsp1 eisosomes.

Heat stress causes a transient increase in the levels of LCBs, phytosphingosine (PHS) and dihydrosphingosine (DHS) (Dickson et al., 1997; Jenkins et al., 1997). In turn, LCB levels regulate Pkh1,2-mediated phosphorylation of Pil1 and Lsp1, which governs the eisosome assembly (Zhang et al., 2004; Walther et al., 2007; Luo et al., 2008). We, therefore, hypothesized, that the increase in LCBs in response to heat stress could be sufficient to trigger the formation of Lsp1 eisosomes in the absence of Pil1. To test this hypothesis, we supplied exponentially growing cells with exogenous PHS. As expected, the effect was comparable to that of heat stress. Specifically, eisosome remnants dissolved and Lsp1 eisosomes were formed in cells incubated with PHS (Fig. 3A,B). Consistent with this, the mean intensity and length of Lsp1–GFP patches in the plasma membrane decreased (Fig. S2).

Cultivation of the cells in the presence of salt represents hyperosmotic shock, leading to a decrease in the tension of the plasma membrane. This has been shown to inhibit the activity of TORC2 (Riggi et al., 2018), an important activator of SPT, which facilitates the first and rate-limiting step of de novo sphingolipid biosynthesis. Consistent with this, sphingolipid catabolism enzymes are upregulated by salt stress in a manner dependent on the high-osmolarity glycogen (HOG) pathway, whereas sphingolipid biosynthesis enzymes are downregulated (Manzanares-Estreder et al., 2017). We therefore tested whether a targeted drop in sphingolipid levels was sufficient to induce the disintegration of eisosome remnants. To induce a general decrease in sphingolipid biosynthesis intermediates, we used the well-described SPT inhibitor myriocin (Miyake et al., 1995; Huang et al., 2012; Tanigawa et al., 2012; Pimentel et al., 2022). In the pil1Δ strain, the decreased SPT activity and resulting sphingolipid insufficiency induced dissolution of remnants and formation of Lsp1 eisosomes, an increase of mean cellular Lsp1–GFP fluorescence and elevation of PM/intracellular GFP intensity ratio (Fig. 3C,D; Movie 1). This indicates an increase in Lsp1 protein abundance, consistent with previously reported overexpression of LSP1 following cultivation in the presence of myriocin (Liu et al., 2013), and increased association of the Lsp1 protein with the plasma membrane. In addition, the mean intensity of the plasma membrane-associated Lsp1–GFP patches increased following myriocin treatment, while they became shorter (Fig. S2). Overall, the effect of myriocin treatment was comparable to that of salt stress, as expected. A decrease in the activity of the SPT can also be achieved genetically by exchanging the native LCB1 gene, coding an essential subunit of the complex, for the temperature-sensitive lcb1-100 allele carrying a missense mutation (Zanolari et al., 2000; Hearn et al., 2003). Analogous to what was seen with myriocin treatment of the pil1ΔLCB1 strain, the non-stressed pil1Δlcb1-100 cells formed Lsp1–GFP eisosomes, whereas their isogenic pil1Δ parent only had eisosome remnants (Fig. 3E; Fig. S2E).

As the addition of PHS and myriocin have opposite effects on the levels of cellular sphingolipids (Pimentel et al., 2022), it is quite curious that both induce the formation of Lsp1 eisosomes in pil1Δ cells. To get more insight into the regulation of this process, and to pinpoint specific sphingolipid pathway intermediates responsible for the changes in Lsp1–GFP localization we used aureobasidin A to inhibit Aur1, an enzyme converting ceramides into inositolphosphoceramides (Heidler and Radding, 1995; Nagiec et al., 1997). The effect of such inhibition is an accumulation of PHS and ceramide (analogous to exogenous PHS addition) and a decrease in the level of complex sphingolipids (as after myriocin treatment) (Nagiec et al., 1997; Schorling et al., 2001; Voynova et al., 2014). Surprisingly, this combined effect of aureobasidin A treatment had no discernible effect on Lsp1–GFP localization in either wild-type cells or the pil1Δ strain (Fig. 3; Fig. S2). This indicates that the formation of Lsp1 eisosomes in the response to stress is not connected to the level of a specific class of sphingolipids.

Therefore, we considered that the addition of either myriocin or PHS, and the use of the lcb1-100 mutant allele all lead to downregulation of the activity of SPT, whereas aureobasidin A treatment does not. Myriocin and lcb1-100 directly decrease the enzymatic activity of SPT (Miyake et al., 1995; Zanolari et al., 2000). Exogenous addition of PHS increases the pool of both LCBs (the products of SPT reaction) and complex sphingolipids (Pimentel et al., 2022), which regulate SPT activity via the MCC/eisosome→TORC2→SPT feedback loop (Fröhlich et al., 2009; Berchtold et al., 2012; Zahumenský et al., 2022). Our data, therefore, indicate a negative correlation between Lsp1 eisosome formation and the activity of SPT.

To uncover the underlying principle of Lsp1 eisosome formation in response to stress, we considered that heat stress (Dunayevich et al., 2018; Winkler et al., 2002), salt stress (Bermejo et al., 2008; Vallejo and Mayinger, 2015) and inhibition of sphingolipid synthesis (Tanigawa et al., 2012) all activate high-osmolarity glycerol (HOG) pathway. We therefore tested whether exposure of pil1Δ cells to high osmolarity induced by 1 M sorbitol, a known activator of the HOG pathway (García-Rodríguez et al., 2005), was sufficient to trigger the formation of Lsp1 eisosomes. Given that sorbitol induces transient phosphorylation of the cell wall integrity (CWI) pathway mitogen-activated protein kinase (MAPK) Slt2 45–60 min after addition (García-Rodríguez et al., 2005), we treated the cultures with sorbitol for 25 min to prevent possible HOG–CWI interference. The changes in Lsp1–GFP localization following sorbitol treatment were comparable to those produced by the cultivation of cells in the presence of 1 M NaCl (Fig. 4A–D; compare with Fig. 1). Specifically, eisosome remnants disintegrated and Lsp1 eisosomes were formed instead. These were distributed almost evenly along the plasma membrane. In the wild-type strain, the number of eisosomes was unaffected by sorbitol treatment (Fig. 4A,B). To completely exclude the possibility that the activation of the CWI pathway contributes to Lsp1 eisosome neogenesis, we treated the cultures with 100 µg/ml Calcofluor White, a known selective activator of the CWI pathway (García-Rodríguez et al., 2005). We found no effect on Lsp1–GFP localization in either pil1Δ cells or the wild-type cells, and conclude that cell wall stress does not induce eisosome remnant disassembly and Lsp1 eisosome formation (Fig. 4A–D).

Given that we discovered Lsp1 eisosome formation in pil1Δ cells in response to hyperosmotic stress, we tested whether the HOG pathway is required for this process by the deletion of the MAPK gene HOG1. This had no significant effect on the Lsp1–GFP localization in either presence or absence of PIL1. Likewise, Lsp1–GFP localization changed in both strains following sorbitol addition in a manner comparable to strains expressing HOG1. Eisosome remnants disassembled and Lsp1 eisosomes were formed (Fig. S3). Our results indicate that the Hog1 MAPK is not required for the formation of Lsp1 eisosomes in response to hyperosmotic stress. If HOG pathway takes part in the process, then the step triggering eisosome formation lies upstream of Hog1.

The eisosome organization is regulated by the Pkh1 and Pkh2 kinase-mediated phosphorylation of Pil1 and Lsp1, which depends on the levels of long-chain sphingoid bases DHS and PHS (Walther et al., 2007; Luo et al., 2008; Baxter et al., 2011). Given that LCBs stimulate phosphorylation of the Lsp1 protein (Zhang et al., 2004), we expected to find decreased Lsp1 phosphorylation under conditions when SPT activity is lowered. Indeed, using a modified western blot protocol, which facilitates the separation of phosphorylated species, we found a decrease in Lsp1 phosphorylation in pil1Δ cells following hyperosmotic stress (Fig. 4E). We conclude that the remodeling of eisosome remnants into Lsp1 eisosomes in response to stress is regulated by phosphorylation of Lsp1.

We found that Lsp1 in S. cerevisiae is able not only to participate in the assembly of the Pil1-organized eisosome but also to assemble the eisosome on its own, in the absence of Pil1, under certain conditions. We asked whether, in the absence of Pil1, Lsp1 was necessary for eisosome formation, and whether Lsp1 alone was sufficient to assemble eisosomes or requires other MCC/eisosome components.

To answer the first question, we followed the localization of Seg1–GFP, an enhancer of eisosome formation and stability (Seger et al., 2011; Moreira et al., 2012; Vaskovicova et al., 2015). In the wild-type strain, Seg1–GFP had a typical eisosome pattern at the plasma membrane (Fig. 5A). In the absence of PIL1, when remnants are formed instead of eisosomes, Seg1–GFP accumulated in eisosome remnants, as was the case for Lsp1 (compare with Figs 1, 3). Its mean cellular fluorescence intensity increased compared to that in the wild type (Fig. 5A,C). The deletion of LSP1, on the other hand, caused no discernible effect on the localization of Seg1–GFP. In the absence of both core eisosome components, Seg1–GFP localized almost exclusively into the cytoplasm with a very limited number of faint plasma membrane-associated patches and eisosome remnants (Fig. 5A,B,D). When cells were cultivated in the presence of 1 M NaCl, there was no change in the Seg1–GFP distribution when Pil1 was present. In the absence of Pil1, LSP1 expression was required to enhance the association of Seg1 with the plasma membrane. In the pil1Δ cells, Seg1–GFP reformed from eisosome remnants into multiple well-defined plasma membrane foci (Fig. 5A,B), where it invariably colocalized with Lsp1 (Fig. 5E). In contrast, in the pil1Δlsp1Δ strain, a change in Seg1 distribution induced by the salt stress was not clearly discernible, mainly because the plasma membrane-associated fraction of the protein remained low. Eisosome remnants could be seen even in stressed cells (Fig. 5A,B). We, therefore, conclude that in the absence of Pil1, Lsp1 is necessary for eisosome assembly during the stress response. The presence of Seg1 foci in the pil1Δlsp1Δ strain is consistent with the idea that recruitment of Seg1 to the plasma membrane during eisosome assembly precedes that of Pil1 and Lsp1 and provides a platform for the assembly of other eisosome components (Moreira et al., 2012).

To answer the question of which proteins are required for eisosome formation in the absence of Pil1, we focused on two proteins known to promote eisosome stability across fungal species, the eisosome constituent Seg1 (Seger et al., 2011; Moreira et al., 2012; Kabeche et al., 2014; Vaskovicova et al., 2015) and the MCC protein Nce102, the absence of which has been shown to significantly reduce the eisosome number (Fröhlich et al., 2009; Loibl et al., 2010; Kabeche et al., 2011; Athanasopoulos et al., 2015). As expected, the deletion of either SEG1 or NCE102 resulted in a decrease, but not complete absence, of Lsp1–GFP foci in both wild-type and the pil1Δ background in the absence of stress (Fig. 6A,B). The mean cellular Lsp1–GFP fluorescence intensity was unaffected by these deletions (Fig. 6C). In the pil1Δ background, salt stress induced the formation of Lsp1 eisosomes in both seg1Δ and nce102Δ mutants, but to a lesser degree than in the strain only lacking PIL1, whereas it did not affect the mean Lsp1–GFP intensity in these strains. In contrast, the mean cellular Lsp1–GFP intensity increased as a result of the simultaneous deletion of SEG1 and NCE102 in the pil1Δ background. However, the formation of Lsp1–GFP foci at the plasma membrane was almost completely abolished and was not induced by salt stress in this strain (Fig. 6). Our results, therefore, indicate that in the absence of Pil1, Lsp1 alone is not sufficient to form eisosomes in response to stress, but requires other proteins, either Seg1 or Nce102.

The physiological role of the eisosome has not yet been fully elucidated. The existence of paralogous genes in the S. cerevisiae genome, which have arisen from whole genome duplication during evolution, has contributed significantly to this failure. In many cases, these homologs either partially or completely substitute for each other, resulting in the masking of phenotypes. Above, we constructed a strain lacking the PIL1, SEG1 and NCE102 genes, which appears to be the minimum requirement for a strain devoid of both eisosomes and eisosome remnants that do not form even when cells are exposed to stress.

To investigate the physiological function of the eisosome under stress, we tested the growth of mutant strains described above under various stress conditions. When grown at the suboptimal 37°C temperature, the viability of the studied strains was not severely affected. Slightly reduced growth was detected in seg1Δ, seg1Δpil1Δ and seg1Δnce102Δ strains (Fig. 7A). When cells were grown in the presence of 1 M NaCl, the absence of either Pil1 or Seg1 decreased the viability of the cells, but the effect was again small. Cultivation of cultures exposed to salt stress at 37°C significantly exacerbated the observed defects. Under conditions of heat stress, the sensitivity of the tested mutant strains to high salt exposure can be summarized as follows: in the absence of Pil1, Nce102 presence causes salt stress sensitivity. In the presence of Pil1, Seg1 confers salt stress resistance. This phenotype is not solely due to misregulation of the sphingolipid synthesis, as the absence of Pil1 induced resistance to 0.5 µg/ml of myriocin independently of NCE102 and SEG1 genes. Increasing the cultivation temperature to 37°C and increasing the myriocin concentration had an additive effect, as expected. This is exemplified well by the similarity of growth of the mutants exposed to 0.5 µg/ml myriocin at 28°C, on one hand, and to 0.25 µg/ml myriocin at 37°C, on the other hand (Fig. 7A). We conclude that Pil1, Seg1 and Nce102 cooperate to coordinate the yeast halotolerance. When all three are absent, the cells are, like wild-type, insensitive to high salt even under conditions of heat stress. To get more insight into the physiological role of the eisosome, we performed a corresponding series of experiments also with the lsp1Δpil1Δ strain, which does not form eisosomes. Unlike the seg1Δnce102Δpil1Δ strain, however, it has eisosome remnants (Fig. 5A). Under all tested conditions, deletion of LSP1 in any background did not change the phenotype of its isogenic parent strain.

Our data indicate that the newly constructed eisosome-less seg1Δnce102Δpil1Δ strain has growth phenotypes that are distinct from both those of the wild type and the pil1Δ strain. To obtain a better understanding of the underlying causes of these differences, we compared the whole transcriptome of the three strains using next-generation sequencing (NGS). To get an initial insight into the data, we performed principal component analysis (PCA), which analyzes correlations among the samples and creates a plot in which samples with similar expression profiles cluster together. Although the biological replicates of the wild-type and pil1Δ strain overlapped, there was a cluster where the seg1Δnce102Δpil1Δ strain clearly separated from the other samples (Fig. 8A). This indicates that although the expression profiles of wild-type and the pil1Δ strain are comparable, there is a significant difference in the transcriptome of the seg1Δnce102Δpil1Δ strain.

Consistently, although in-depth analysis reported no differentially expressed genes (DEGs) (|log₂FoldChange|>1; adjusted P<0.05) in the pil1Δ strain, there were 1069 DEGs in the triple deletion mutant when compared to the wild type. A total of 635 of these genes were downregulated, whereas 434 genes, including LSP1 (consistent with the mean cellular GFP intensity increase shown in Fig. 6A,C), were upregulated (Fig. 8B,C; Table S3). Performing Gene Set Enrichment Analysis (GSEA) on KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways revealed significant changes (adjusted P<0.05) in 15 pathways/processes (Fig. 8D,E). The most affected were oxidative phosphorylation and tricarboxylic acid cycle (upregulated), and DNA replication, ribosome biogenesis and cell cycle (downregulated).

The core eisosome proteins Pil1 and Lsp1 share ∼72% sequence identity. However, their contribution to the assembly of the hemitubular scaffold of the eisosome is remarkably different. Under normal laboratory conditions, eisosomes collapse into eisosome remnants when Pil1 is not present in the cell. The absence of Lsp1, on the other hand, has no apparent effect on eisosome morphogenesis (Walther et al., 2006; Grossmann et al., 2007; Olivera-Couto et al., 2011; Deng et al., 2009). It was therefore believed that, on its own, Lsp1 was insufficient to organize the eisosomes. In our present study, we report that under certain conditions, such as hyperosmotic stress, the eisosome remnants of pil1Δ cells disintegrate and eisosomes are formed even without Pil1. These Lsp1 eisosomes require the integral MCC protein Nce102 and the eisosome component Seg1 for their formation.

It has been previously reported that hyperosmotic stress, which induces cell shrinkage, induces morphological changes in the MCC/eisosome microdomain. The resulting decrease in membrane tension promotes deepening of eisosomes (Schaber et al., 2010; Dupont et al., 2010; Appadurai et al., 2020) and the accumulation of phosphatidylinositol 4,5-bisphosphate (PIP₂) in the furrows (Riggi et al., 2018). In the absence of Pil1, the distribution of PIP₂ in the plasma membrane changes from being quite punctate to being more homogeneous, and the lipid becomes more accessible to the binding of the GFP–2xPH_PLC probe (Fröhlich et al., 2014). Hyperosmotic stress induces the formation of PIP₂ clusters in the plasma membrane (Kabeche et al., 2015; Riggi et al., 2018). It is plausible that these clusters provide a binding platform for Seg1 [via its BAR and PH domains (Moreira et al., 2012)] and Lsp1 (via the BAR domain; Ziółkowska et al., 2011), and that the interaction of these proteins further enhances their binding to the plasma membrane. This points to the general mechanism of eisosome biogenesis. Correspondingly, we show that heat stress is another inducer of Lsp1 eisosome formation in the absence of Pil1. Heat stress has been shown to stimulate a transient increase of cellular PIP₂ (Desrivières et al., 1998). The transiency of this change might explain why heat stress is a weaker inducer of Lsp1 eisosome formation than salt or hyperosmotic stress. Further support for the PIP₂-based mechanism of eisosome formation comes from our observation that under salt stress conditions, the plasma membrane association of Seg1 slightly increases even in the strain lacking both eisosome organizers, Pil1 and Lsp1 (Fig. 5).

Pil1, Nce102 and other proteins are known to leave the MCC/eisosome when the cellular need for sphingolipids is not met (Fröhlich et al., 2009; Zahumenský et al., 2022), leading to destabilization of the whole microdomain. By contrast, Lsp1 is required for elongation of eisosomes and an increase in their number in the stationary phase (Gournas et al., 2018), that is, in a respiring culture with limited nutrient supply. It should be emphasized here that our observation of increased eisosome assembly in myriocin-treated pil1Δ cells (Fig. 3) is not contradictory to the earlier report of partial eisosome disassembly following myriocin treatment in the wild-type cells (Fröhlich et al., 2009). The authors of the cited study identified the release of Nce102 as the primary reason for the decrease in the eisosome number. We showed recently that this occurs specifically during cytokinesis, which the cells undergo in the presence of myriocin (Zahumenský et al., 2022). A decreased amount of Nce102 in the MCC/eisosome, reporting here the local lack of sphingolipids, mimics the situation in the nce102Δ mutant, which exhibits a decreased eisosome number (Fröhlich et al., 2009). The number of Lsp1 eisosomes formed in the absence of Pil1 under the stress conditions or in myriocin-treated cells (Figs 1 and 3) is significantly lower compared to that in the wild-type cells. It could be that Lsp1-stabilized eisosomes are those which persisted after the myriocin treatment in both cases, either in the wild-type or pil1Δ strain. If so, the only difference would be that, in the absence of Pil1, Lsp1 eisosomes are newly formed in the plasma membrane, while in the presence of Pil1, dissociation of Pil1 and increased association of Lsp1 with the plasma membrane following the myriocin treatment would occur together.

We report that the formation of Lsp1 eisosomes in response to stress requires the presence of the integral MCC protein Nce102 and the soluble eisosome component Seg1. The Lsp1 eisosomes are less abundant in the absence of either and all but completely absent when neither is present. Our microscopy (Fig. 6C) and NGS (Table S1) data indicate an increase in the level of Lsp1 in the seg1Δnce102Δpil1Δ strain, which suggests that the cell increases the production of available eisosome components in the attempt to build the structure. Consistent with this, our NGS data indicate downregulation of the MAK3 gene, deletion of which in otherwise wild-type cells leads to over-assembly of eisosomes and their aberrant elongation (Grossmann et al., 2008; Malinsky et al., 2010). Our NGS analysis reported that there are no significant differences in the transcriptome of the pil1Δ strain compared to wild type. This corresponds very well with the fact that this strain exhibits only weak phenotypes when cultures are grown under normal laboratory conditions.

Analyzing the seg1Δnce102Δpil1Δ strain, the plasma membrane of which is devoid of eisosomes under both normal and stress conditions, we found that its stress resistance differs from both the wild-type cells and the pil1Δ strain. Like the pil1Δ, the triple mutant shows increased resistance to myriocin compared to the wild type. Unlike the pil1Δ, however, the halotolerance of the triple mutant does not differ from the wild type. In a recent study, a different eisosome-lacking strain has been described (Sakata et al., 2022). In this strain, genes coding both Nce102-like proteins (Nce102 and Fhn1) and four MCC-localizing members of the Sur7 family (Sur7, Pun1, Fmp45 and Ynl194c) were removed in addition to PIL1. The resulting strain had elevated sensitivity to salt stress and the membrane-disruptive chemical SDS, which the authors ascribed to hyperactive TORC2 signaling. Our simpler eisosome-lacking mutant, and also the nce102Δpil1Δ strain, showed no sensitivity to either salt stress (Fig. 7A) nor SDS (data not shown). Keeping in mind low FHN1 expression levels under given conditions (Loibl et al., 2010), this difference indicates a possible protective role of Sur7 family proteins in salt stress.

We found that the Hog1 MAPK is not required for the formation of Lsp1 eisosomes in pil1Δ cells in response to hyperosmotic stress. This contrasts with the known existence of a functional connection between the eisosome-regulated CWI pathway and the eisosome component Pil1. Specifically, the CWI MAPK Slt2 has been shown to phosphorylate Pil1, and thus provide a feedback loop between the pathway activation and eisosome assembly (Mascaraque et al., 2013). On the other hand, the absence of the MAPK of the HOG pathway does not impair the eisosome assembly, indicating that another (upstream) step of the pathway possesses an eisosome-related regulatory function. Further investigation will be necessary to identify this regulator. At the moment, we can only speculate about the role of the Sho1 branch of the HOG pathway, which includes lateral rearrangement (clustering) of the receptor components in the plasma membrane upon activation (Tatebayashi et al., 2007, 2015).

Previously, the MCC/eisosome has been linked to the regulation of nutrient transporter turnover (Grossmann et al., 2007; Gournas et al., 2017; Busto et al., 2018), RNA degradation (Grousl et al., 2015; Vaškovičová et al., 2017) and sphingolipid biosynthesis (Fröhlich et al., 2009; Berchtold et al., 2012; Zahumenský et al., 2022). The construction of a strain that forms neither eisosomes nor eisosome remnants allowed us to identify novel functional connections between the MCC/eisosome microdomain and essential cellular processes. Transcriptome analysis of this strain revealed significant changes in the metabolism of lipids, oxidative phosphorylation, MAPK signaling and cell cycle progression, clearly indicating novel avenues to be explored in future eisosome research.

S. cerevisiae strains used in this study are listed in Table S1. Yeast cells were incubated in a complete synthetic (SC; 0.17% yeast nitrogen base without amino acids and ammonium sulfate, 0.5% ammonium sulfate, 2% glucose, amino acids) or rich medium (YPD; 2% peptone, 1% yeast extract, 2% glucose) at 28°C on a shaker. In all experiments including the lcb1-100 allele-containing strain, the cultivation was performed at 25°C. For assessment of the effects of chemical stress on Lsp1–GFP localization, cultures were grown in SC medium as described above for 4 h, followed by the addition of aureobasidin A (1 mg/ml ethanol stock), myriocin (2 mg/ml methanol stock; Sigma, M1177), or phytosphingosine (PHS, 5 mg/ml ethanol stock; Sigma, P2795) to desired concentrations and additional 2-h cultivation. In the case of salt stress, cells were cultivated in the presence of 1 M NaCl for 6 h. Hyperosmotic stress and cell wall stress were induced by cultivating the cells for 6 h in SC, followed by their 25-min exposure to 1 M sorbitol (added as 0.91 g powder to 5 ml of culture) or 100 µg/ml Calcofluor White (1 mg/ml stock), respectively. All chemicals were purchased from Merck. For heat stress, cells were grown for 4 h as described above, then shifted from 28°C to 37°C. For electron microscopy preparations, the cells were cultured in YPD. For propagation of plasmids, Escherichia coli strain XL1-Blue (Stratagene, San Diego, CA, USA) was used. Bacterial strains were incubated in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl) supplemented with ampicillin (100 µg/ml) for selection of transformants.

Construction of the YIp128-SEG1-GFP plasmid has been described previously (Vaskovicova et al., 2015). The YIp128-LSP1-GFP plasmid was constructed as follows. The gene of interest was inserted as a HindIII-BamHI fragment into the YIp128-GFP plasmid. Before the transformation, the plasmid was linearized by digestion with the PmlI restriction enzyme. The pKT128-LSP1-GFP cassette was amplified by PCR from the pKT128 plasmid (EUROSCARF, P30174) using the primers LSP1-gnmtag_F, 5′-ATGTTAGTCAAAACGGTCATACCTCCGGTTCTGAAAACATCCGGATCGGTGACGGTGCTGG-3′ and LSP1-gnmtag_R, 5′-TAGGTAGACACGAGCTGGTGAAAAACAAGCACAATATAGAAAGGTTCGCATCGATGAATTCGAGCTCG-3′. The obtained fragment was used to transform the yeast BY4742 strain (EUROSCARF). Candidate colonies were selected on histidine drop-out plates. The pCM225-LSP1-GFP cassette was amplified by PCR from the pCM225 plasmid (EUROSCARF, P30342) using the primers LSP1_pCM225_ctrl_F, AAGATGCAACGGCTAAATCGCAATATATAAACAGGTGAGATAGCAGCTGAAGCTTCGTACGC-3′ and LSP1_pCM225_ctrl_R, 5′-AGCCGTTGGAGCTCTTTGATTTCTTAAAGAGTAAGTTCTGTGCATATAGGCCACTAGTGGATCTG-3′. The obtained fragment was used to transform the yeast BY4742 strain (EUROSCARF). Candidate colonies were selected on YPD supplemented with kanamycin G418 (100 µg/ml; Gibco, 11815-024) for the selection of transformants. The pFA6a-natMX6_PIL1 deletion cassette was amplified by PCR from the pFA6a-natMX6 plasmid (EUROSCARF, P30437) using the primers PIL1_del_pFA6a_F, 5′-TATTGCAAAGTGAAGAATATATCAGCATCAAGTATATAGTCGGATCCCCGGGTTAATTAA-3′ and PIL1_del_pFA6a_R, 5′-TTAATTAAAAGAGATTAATTAGTATGGTAACTTGTTCTTTTCTGCTGGGAATTCGAGCTCGTTTAAAC-3′. The obtained fragment was transformed into strains Y1214, Y1233, Y1234 and Y1417 (Table S1). Subsequently, the transformants were selected on YPD supplemented with nourseothricin (100 µg/ml; Jena Bioscience AB102L). The presence of the pFA6a-natMX6_PIL1 deletion cassette was verified by colony PCR.

The double- and triple-deletion strains were obtained by mating suitable single- and double-deletion strains (Table S2) and subsequent sporulation and tetrad dissection. Mutants were selected with the help of the fluorescence microscopy as non-fluorescing colonies. Successful deletions were then verified by PCR.

Living yeast cells grown in SC medium were concentrated by a brief centrifugation (1430 g for 2 min), immobilized on a 0.17 mM cover glass by a thin film of 1% agarose prepared in 50 mM potassium phosphate buffer (pH 6.3) and imaged using LSM 880 (Zeiss; 100× PlanApochromat oil-immersion objective; NA=1.4) laser scanning confocal microscope. The fluorescence signal of GFP (excitation 488 nm; Ar laser) and mRFP/mCherry (excitation 561 nm; solid-state laser) was detected using the 493–550 nm and 578–696 nm band-pass emission filters (Zeiss), respectively.

Cells from an overnight culture were harvested by brief centrifugation (1 min at 1500 g) and washed in 50 mM potassium phosphate buffer (pH 5.5). A 2 μl aliquot of the concentrated cell suspension was loaded onto a gold carrier and frozen rapidly in liquid nitrogen. The sample was cut with a cold knife (T≤−185°C), etched for 4 min (−97°C; pressure≤1.3105 Pa) in a CFE-50 freeze-etch unit (Cressington, Watford, UK), shadowed (1 nm Pt/C, 45°C; 10 nm C, 90°C), and cleaned in fresh 70% H₂SO₄ for 16 h. Air-dried sample surface replicas were examined using an FEI Morgagni 268(D) transmission electron microscope at 80 kV. Images were captured with a MegaView II CCD camera (Olympus Corp., Münster, Germany).

Yeast cells were grown for 6 h and harvested by centrifugation (1430 g for 2 min). The pellet was resuspended in distilled water, precipitated by the addition of 0.6 ml of 100% (w/v) tricarboxylic acid (TCA) and kept on ice for 15 min before being centrifuged (5 min, 2460 g, 4°C). The supernatant was aspirated and the remaining pellet was washed twice with cold acetone. After spinning (5 min, 2460 g, 4°C) and removal of residual medium the pellet was dried for 60 min at room temperature. 300 µl of urea buffer was added [50 mM HEPES pH 7.4; 8 M urea; 2% SDS; cOmplete^TM protease inhibitor cocktail (Roche, Basel, Switzerland); 10 mM PMSF; 10 mM ABA, 100 µg/ml leupeptin and 100 µg/ml pepstatin (all Sigma-Aldrich, St Louis, MO, USA)], and resuspended. For monitoring protein phosphorylation, NEB lysis buffer [50 mM HEPES pH 7.5, 100 mM NaCl, 2 mM DTT, 1% Triton^TM X-100, 1 mM MnCl₂; cOmplete^TM protease inhibitor cocktail and phosphatase inhibitor cocktail 3 (except for the λPPase sample)] was used. Cells were homogenized by beating (5×45 s, 4000 rpm) with glass beads (0.32-0.43 mm) in a BeadBug microtube homogenizer (Benchmark Scientific, Edison, NJ, USA). For dephosphorylation, the reaction mixture was then supplemented with 400 units of λPPase (NEB), and the reaction was allowed to proceed for 30 min at 30°C according to the manufacturer's instructions. Samples were diluted with respective buffers [based on Pierce^TM BCA protein assay (Thermo Fisher Scientific) or direct protein concentration measurement using a NanoDrop^TM ND-1000 Spectrophotometer for the phosphorylation study], mixed with 5× Laemmli protein sample buffer (312.5 mM Tris-HCl pH 6.8, 10% SDS, 25% glycerin, 25% β-mercaptoethanol and 0.2% Bromophenol Blue) and heated at 95°C for 5 min. Proteins were resolved using a 12.5% SDS-polyacrylamide gel or 6% 20 μM Zn(II) Phos-Tag^TM SDS-polyacrylamide gel (40 μg of total protein in each sample) and wet-transferred to an Immobilon-E PVDF membrane (0.45 µm pore, Merck Millipore, Burlington, MA, USA).

The membrane was blocked with 5% milk (Serva Electrophoresis, Heidelberg, Germany) in TBST buffer (50 mM Tris-HCl pH 7.4 150 mM NaCl and 0.05% Tween 20) for 60 min, and incubated for 4 h at room temperature in 1% milk in TBST buffer with primary antibodies. To detect Lsp1–GFP, a mixture of anti-GFP mouse monoclonal conjugated with horseradish peroxidase (HRP; 1:1000, cat. no. sc-9996, Santa Cruz Biotechnology, Dallas, TX, USA) and anti-tubulin (YL1/2) rat monoclonal (1:10,000, ab6160, Abcam, Cambridge, UK; loading control) was used. After washing, the membrane was imaged for anti-GFP signal and subsequently incubated for 60 min at room temperature in 1% milk in TBST buffer with secondary antibody: for tubulin, anti-rat-IgG (goat, 1:10,000, Jackson ImmunoResearch, West Grove, PA, USA); for Lsp1–GFP, none. HRP chemiluminescence was monitored with Azure c400 (Azure Biosystems, Dublin, CA, USA) and VWR^® Imager CHEMI Premium (VWR International, Radnor, PA, USA) detection system and analyzed using Image Studio Lite Ver 5.2 and VWR^® Image Capture Software.

Sensitivity to a range of stresses was evaluated using a solid medium assay. All strains were grown overnight in a rich YPD medium at 28°C, then diluted to equivalent concentrations of am optical density at 600 nm (OD₆₀₀)=0.5 in water, and 10 μl of a 3-fold dilution series were spotted with a replica plater (R2383, Merck) onto YPD agar plates (2% agar, 2% peptone, 1% yeast extract, 2% glucose) with or without these chemicals: NaCl (1 M), myriocin (0.25 ug/ml or 0.5 ug/ml). Cells were grown on plates for 2–3 days at 28°C or 37°C, imaged with a Canon CanoScan 5600F scanner and growth differences were evaluated.

Indicated strains were cultivated in 10 ml of SC medium for 24 h, pelleted (2460 g for 4 min, 4°C) and washed with 10 mM NaF/NaN₃. The supernatant was aspirated and the cells were homogenized by beating (5×30 s, 4000 rpm) with glass beads (0.32–0.43 mm) in 600 µl RLT buffer (RNeasy^® kit, Qiagen) in a BeadBug microtube homogenizer (Benchmark Scientific, Edison, NJ, USA) and pelleted (610 g for 1 min). The supernatant was replaced into a fresh tube and treated with 10 μl proteinase K (Thermo Fisher Scientific) for 10 min at room temperature. Total RNA was isolated using the RNeasy^® Plus Mini Kit (includes gDNA removal columns) followed by DNA digestion with a DNase Max^® Kit (both Qiagen), according to the manufacturer's protocols. Samples were frozen in liquid nitrogen and delivered to the sequencing facility. RNA quality number (RQN, Agilent Fragment Analyzer 5200) of all samples was greater than 8. Sequencing libraries were prepared using polyA selection and sequenced using Illumina NextSeq 500.

Sequencing data were analyzed in R Studio. Quality of sequencing was verified using FastQC, rRNA and tRNA removed using SortMeRNA (Kopylova et al., 2012), and reads aligned to S288C genome reference (version R64-3-1_20210421; Saccharomyces genome database) using GSNAP (Wu and Nacu, 2010). Differential gene expression was analyzed using the DESeq2 package (Love et al., 2014), including normalization (median of ratios method) and log fold change (LFC) shrinkage (Stephens, 2017), with the significance cut-off adjusted P-values (corresponding to false discovery rate)<0.05 and LFC (|log₂FoldChange|)>1. Data visualization was performed using the ggplot2 package (Wickham, 2016). For exploratory analysis (PCA, GSEA and hierarchical clustering), normalized sequencing counts were further rlog2 transformed.

Image processing and analysis were performed in Fiji (ImageJ 1.53c) using custom-written macros, available at https://github.com/jakubzahumensky/Nce102_SL_paper (Zahumenský et al., 2022) and https://github.com/jakubzahumensky/Isc1_paper (Balazova et al., 2022) and cell segmentation masks made using the Cellpose software (Stringer et al., 2021), as described previously (Zahumenský et al., 2022). Statistical analysis was performed in GraphPad Prism 9 software. Statistical significance was determined by two-way ANOVA with Dunnett post test. Data distribution was assumed to be normal, but this was not formally tested.

NGS was performed at the Sequencing facility of IMG CAS on a commercial basis. Analysis of NGS data was greatly facilitated by R scripts and RMarkdown documents previously developed by Jiří Novotný from IMG CAS. Special thanks belong to Satyendra Mondal who isolated the yeast RNA used for NGS analysis.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260554.reviewer-comments.pdf