Distinct roles for plasma membrane PtdIns(4)P and PtdIns(4,5)P₂ during receptor-mediated endocytosis in yeast

Clathrin-mediated endocytosis is the main pathway for internalization of membrane proteins, such as cell-surface receptors and transporters. In the budding yeast Saccharomyces cerevisiae, nearly 60 proteins are recruited and activated at distinct sites on the plasma membrane to regulate endocytosis (Boettner et al., 2011; Weinberg and Drubin, 2012). Based on their functions and the timing of their appearance and disappearance at endocytic sites, these endocytic proteins are divided into several groups: early proteins, early, middle and late coat proteins, WASp and myosin-related proteins, actin and actin-associated proteins, and scission-related proteins (Kaksonen et al., 2005; Lu et al., 2016). In the class of early or middle coat proteins, several proteins contain a lipid-binding domain. The yeast epsins Ent1 and Ent2 (hereafter referred to as Ent1p and Ent2p, respectively), contain a highly conserved epsin-N-terminal homology (ENTH) domain that binds to phosphorylated derivatives of phosphatidylinositol (Itoh et al., 2001). In addition, the yeast AP180 clathrin adaptor proteins, Yap1801p and Yap1802p, and the yeast Hip1R homolog Sla2p contain a AP180 N-terminal homology (ANTH) domain, the module highly related to the ENTH domain (Ford et al., 2001), and interaction of Sla2p's ANTH domain with phosphatidylinositol-4,5-bisphosphate [PtdIns(4,5)P₂] has been shown to be important for actin-dependent endocytic internalization (Sun et al., 2005). Epsin is targeted to areas of endocytosis and directly modifies membrane curvature by binding to PtdIns(4,5)P₂ in conjunction with clathrin polymerization (Ford et al., 2002). A recent study showed that the membrane-binding domains of Sla2p and Ent1p assemble together in a PtdIns(4,5)P₂-dependent manner to form an oligomeric structure on membranes, which is essential for membrane remodeling during endocytosis (Skruzny et al., 2015). Additionally, PtdIns(4,5)P₂ metabolism through the yeast synaptojanin-like proteins, Sjl1p/Sjl2p/Sjl3p, coordinately directs membrane invagination and the release of endocytic proteins after vesicle scission (Stefan et al., 2005; Sun et al., 2007).

PtdIns(4,5)P₂ also has been shown to bind to actin regulators, and control assembly and turnover of actin filaments (Yin and Janmey, 2003). Binding of PtdIns(4,5)P₂ to profilin, cofilin, or capping protein rapidly causes dissociation of these proteins from actin (Lassing and Lindberg, 1985; Schafer et al., 1996; Yonezawa et al., 1990). In yeast endocytosis, these actin regulators, profilin (Pfy1p), cofilin (Cof1p), and capping proteins (Cap1p, Cap2p), are localized at the endocytic sites and required for clathrin-coated vesicle internalization (Goode et al., 2015). In addition, recent evidence suggests that PtdIns(4,5)P₂ can relieve myosin-I Myo5 autoinhibition, leading to Arp2/3-mediated actin nucleation (Fernández-Golbano et al., 2014). Thus, PtdIns(4,5)P₂ appears to regulate multiple stages of endocytosis by interacting with various endocytic proteins and actin regulators.

In contrast to PtdIns(4,5)P₂, the specific functions of plasma membrane phosphatidylinositol-4-phosphate [PtdIns(4)P] in endocytosis are not established. PtdIns(4)P residing in the plasma membrane has been regarded simply as a precursor of PtdIns(4,5)P₂. While a main route of PtdIns(4,5)P₂ synthesis is through phosphorylation of PtdIns(4)P by PtdIns(4)P 5-kinase, several studies have shown that inhibitors of PI 4-kinase activity, such as LY294002 and phenylarsine oxide (PAO), cause depletion of cellular PtdIns(4)P, but have only minor effects on the total amount of PtdIns(4,5)P₂ (Hammond et al., 2009, 2012). Hammond et al. further demonstrated that most plasma membrane PtdIns(4)P is not required for the synthesis and functions of PtdIns(4,5)P₂ but is, indeed, required for plasma membrane targeting of proteins through basic amino acid stretches that interact with anionic lipid (Hammond et al., 2012). PtdIns(4)P, therefore, might not simply serve as the immediate precursor of PtdIns(4,5)P₂ but, rather, contributes to the pool of polyanionic lipids that define plasma membrane identity.

PI kinases and phosphatases required for the synthesis and metabolism of PtdIns(4,5)P₂ were reported to contribute to endocytosis in yeast cells (Audhya et al., 2000; Singer-Kruger et al., 1998). Mss4p is a functional homolog of mammalian PtdIns(4)P 5-kinase and essential for cell growth (Desrivières et al., 1998; Homma et al., 1998). Previous studies demonstrated that Mss4p is required for proper organization of the actin cytoskeleton (Desrivières et al., 1998; Homma et al., 1998) and the internalization step of endocytosis (Desrivières et al., 2002; Sun and Drubin, 2012). Live-cell imaging of endocytic protein dynamics in an mss4 temperature-sensitive (ts) mutant with severely reduced PtdIns(4,5)P₂ levels, revealed that PtdIns(4,5)P₂ is required for endocytic membrane invagination but is less important for endocytic site initiation (Sun and Drubin, 2012). The majority of PtdIns(4)P in budding yeast is generated by two essential PI 4-kinases encoded by the STT4 and PIK1 genes (Flanagan et al., 1993; Yoshida et al., 1994). Pik1p is localized to the trans-Golgi membrane, and a pik1 ts-mutant accumulates in Golgi structures, and a severe defect in Golgi-to-plasma membrane and Golgi-to-vacuole trafficking, suggesting an important role for Pik1p at the Golgi complex (Audhya et al., 2000; Hama et al., 1999; Walch-Solimena and Novick, 1999). Stt4p localizes to the plasma membrane by forming a complex with Ypp1p and Efr3p (Baird et al., 2008). Several studies have reported that Stt4p is indirectly implicated in regulation of the cell cycle, lipid metabolism, actin cytoskeleton dynamics, and activation of the MAPK pathway (Audhya and Emr, 2002; Audhya et al., 2000; Muhua et al., 1998; Tabuchi et al., 2006). However, the physiological relevance and function of plasma membrane PtdIns(4)P generated by Stt4p has, so far, not been well elucidated.

In this study, we examined the potential independent role for PtdIns(4)P during endocytosis by generating ts-mutants of the key enzymes involved in the life cycle of this lipid: a mss4 mutant to increase the level of PtdIns(4)P, while decreasing the level of PtdIns(4,5)P₂; and the stt4 pik1 double mutant to decrease both PtdIns(4)P and PtdIns(4,5)P₂. Interestingly, we found that recruitment of the ENTH/ANTH-domain proteins, Ent1p, Ent2p, Yap1801p and Yap1802p, to endocytic sites is significantly inhibited by depletion of PtdIns(4)P in the stt4^ts pik1^ts mutant, but is increased in the mss4^ts mutant. We also demonstrated that the stt4^ts pik1^ts mutant exhibits a remarkable delay in endocytic cargo recruitment to clathrin-coated pits, whereas the mss4^ts mutant does not. Therefore, our results suggest that PtdIns(4)P does not simply serve as an immediate precursor to PtdIns(4,5)P₂ but plays a specific role in receptor-mediated endocytosis.

To examine the in vivo role of Stt4p−one of the two essential PI 4-kinases−in receptor-mediated endocytosis, we utilized a stt4 decreased abundance by mRNA perturbation (DAmP) strain (stt4 kd), in which the 3′UTR of the STT4 gene had been disrupted in order to destabilize the transcript (Breslow et al., 2008). In addition to the stt4 kd alleles, we also generated ts kinase alleles, including a pik1 mutant (see below). We first confirmed that expression of Stt4p was decreased to ∼46.4% in the stt4 kd mutant compared with that in wild-type cells (Fig. 1A, left). The stt4 kd cells displayed a growth defect at 25°C and lethality at 37°C, indicating that this mutant is temperature sensitive (Fig. 1B). We next examined localization of PtdInsP isoforms by monitoring fluorescent probes for specific PtdInsPs. To distinguish between wild-type and mutant cells, the former were labeled with Abp1-mCherry (Fig. 1C) or FM4-64 (Fig. 1E). Localization of PtdIns(3)P at the vacuolar membrane was detected by using FYVE-GFP (Burd and Emr, 1998) in wild-type cells; knockdown of Stt4p did not affect localization of this PtdIns(3)P sensor (data not shown). In wild-type cells, we observed that the PtdIns(4)P probe GFP-PH^Osh2, comprising two tandem repeats of the PH domain from Osh2 fused to GFP, was localized at the plasma membrane and intracellular puncta (Fig. 1C, upper left panel), consistent with a previous report (Balla et al., 2008; Roy and Levine, 2004). In wild-type cells, these GFP-PH^Osh2-labeled intracellular puncta (labeled by using Sec7-mCherry) mostly localized to the Golgi-complex (Fig. S1). In the stt4 kd mutant, the plasma membrane localization of GFP-PH^Osh2 decreased to ∼36.2% of that in wild-type cells (Fig. 1C,D), while the localization to the Golgi complex was not affected (data not shown), indicating that Stt4p is required for PtdIns(4)P production at the plasma membrane.

We sought to examine whether Stt4p acts mainly to aid in the synthesis of PI(4,5)P₂ by providing 1-phosphatidylinositol-4-phosphate 5-kinase [PtdIns(4) 5-kinase] with sufficient PtdIns(4)P as its substrate. However, we did not detect a change in the plasma membrane localization of PtdIns(4,5)P₂ (labeled by using GFP-PH^PLCδ) in the stt4 kd mutant (Stefan et al., 2002) (Fig. 1C,F). To explore the role of Stt4p in PI(4,5)P₂ production further, we utilized a DAmP strain of the MSS4 gene, which encodes a PtdIns(4)P 5-kinase that synthesizes PI(4,5)P₂. Expression of Mss4p was decreased to ∼48.7% in the mss4 kd mutant compared with that in wild-type cells (Fig. 1A, right). Consistent with this result, plasma membrane localization of PtdIns(4,5)P₂ decreased to ∼68.5% of that seen in wild-type cells (Fig. 1E,F), but this mutant grew as well as wild type, with no defect detectable at 25°C and 37°C, suggesting that depletion of PtdIns(4,5)P₂ in the mss4 kd mutant does not cause strong effects on cell growth and endocytosis (see Fig. 1G,H). In contrast, the mss4 kd mutant caused severe growth defects when combined with the stt4 kd mutant (Fig. 1B). To test whehter this was due to lower levels of PtdIns(4,5)P₂ at the plasma membrane in the double mutant, we again used GFP-PH^PLCδ as a sensor. In the stt4 mss4 double-knockdown (stt4 mss4 kd) mutant, the plasma membrane localization of PtdIns(4,5)P₂ was only slightly decreased (∼47.2%), compared to 68.5% seen in the mss4 kd mutant. These results suggest that Stt4p does not play a significant role in setting the levels of PtdIns(4,5)P₂ at the plasma membrane and, instead, affects PtdIns(4)P levels there. PtdIns(4)P levels at the plasma membrane, however, increased by ∼1.6- and ∼1.5-fold in the mss4 kd and stt4 mss4 kd mutants, respectively (Fig. 1D-F), suggesting that, in the double mutant, inhibition of PtdIns(4,5)P₂ synthesis at the plasma membrane through the mss4 kd overcompensates for the decrease in PtdIns(4)P caused by the mutation in the PI 4-kinase Stt4p.

We wished to determine the role of PtdIns(4)P in receptor-mediated endocytosis. We first examined endocytosis in the Stt4p mutant by assessing the internalization and transport of yeast mating pheromone α-factor labeled with Alexa Fluor 594 (hereafter referred to as A594-α-factor) (Toshima et al., 2006). As reported previously, when added to wild-type cells, A594-α-factor is first localized to the plasma membrane and then transported to the vacuole, including prevacuolar compartments, within 20 min (Fig. 1G) (Toshima et al., 2006). In the stt4 kd mutant, A594-α-factor was still partially localized at the plasma membrane 20 min after internalization, indicating that this mutant has a slight defect in α-factor internalization. In contrast, the mss4 kd mutant showed little effect on α-factor internalization or transport to the vacuole (Fig. 1G,H). Interestingly, we found that the stt4 mss4 kd mutant has a severe defect in α-factor internalization (Fig. 1G). Quantitative analysis that distinguished between A594-α-factor localization at both plasma membrane and vacuole and vacuole-only localization revealed that the stt4 kd and the stt4 mss4 kd mutant have obvious defects in A594-α-factor transport (Fig. 1H). As for the additive defect in the stt4 mss4 kd mutant, we propose two possibilities. One is that both Stt4p and Mss4p contribute to produce PtdIns(4,5)P₂, and the further reduction in the level of PtdIns(4,5)P₂ (Fig. 1F) led to an additive defect; another possibility is that these proteins function to regulate distinct lipids that are needed for different steps of receptor-mediated endocytosis.

To address the function of Stt4p and Mss4p as well as the lipids they regulate in the endocytic pathway, we generated ts-mutants through PCR-based random mutagenesis of the catalytic domain of these kinases. To this end, we developed a new method to screen ts mutants (see Materials and Methods for details). In brief, to isolate the stt4 ts mutant, we created a plasmid library containing STT4 gene fragments that carry various mutations introduced by an error-prone PCR method (Fig. 2A). To integrate the plasmid library at the endogenous locus of the STT4 gene, the plasmid was linearized by NruI and transformed into wild-type cells; transformants were subsequently grown on SD-URA plates at 25°C or 37°C. Among 12 stt4 ts candidate mutant strains, the stt4-1 strain carrying a Phe1777Ser mutation (Fig. 2B) was selected for further study. By using the same method, we also succeeded in isolating a ts allele of MSS4 (mss4-1) with an Ala660Ser mutation (Fig. 2B). We did the same for the trans-Golgi membrane-localized PI 4-kinase, PIK1 (pik1-1), isolating a ts mutation with an Asp1055Gly change (Fig. 2B). All these mutants exhibited a temperature-sensitive growth phenotype, but their growth rates at 25°C or 37°C were different from one another (Fig. 2C). We first examined the expression of these proteins in mutant cells. As shown in Fig. 2D, expression levels of Stt4p were almost the same between wild-type and stt4-1 cells at 25°C and 37°C, those of Pik1p decreased in pik1-1 cells to ∼14.8% of wild type at 37°C, and those of Mss4p decreased in mss4-1 cells to ∼75.7% of wild type at 25°C and ∼47.2% at 37°C. The localization of these mutant proteins seems to be similar to that of the wild-type proteins, both at 25°C and 37°C (Fig. S2), suggesting that the point mutation in Stt4p makes it nonfunctional, and that in Pik1p or Mss4p makes it unstable.

We next examined the localization of PtdIns(4)P at the plasma membrane and the Golgi complex of these mutant cells. PtdIns(4)P levels at the Golgi were determined by measuring the signal intensity of GFP-PH^Osh2 colocalizing with Sec7-mCH (Fig. S1). In the stt4-1 mutant at the restrictive temperature, the plasma membrane localization of the PI(4)P-binding protein GFP-PH^Osh2 decreased to ∼18.1% of that seen in the wild type, whereas the Golgi localization was not significantly affected (Fig. 3A,B). The similarity of these effects compared with those seen in the stt4 kd suggests that the stt4-1 mutant exhibits a loss-of-function phenotype. In contrast, in the pik1-1 mutant, Golgi localization of the PI(4)P-binding protein GFP-PH^Osh2 decreased to ∼6.0% at the restrictive temperature but the plasma membrane localization was not affected (Fig. 3A,B). Since a previous study had reported that a stt4 pik1 double mutant exhibits an additive effect on growth and PtdIns(4)P production in vivo (Audhya et al., 2000), we created a mutant that harbors both the stt4-1 and pik1-1 alleles (stt4-1 pik1-1). This strain grew more slowly than either single mutant at 25°C (Fig. 2C), and GFP-PH^Osh2 levels at the plasma membrane and in the Golgi was significantly decreased (∼7.1% and ∼6.2%, respectively) at 37°C (Fig. 3A,B). As in the stt4 mss4 kd mutant, we found that the plasma membrane localization of GFP-PH^Osh2 increases ∼2.4-fold in the mss4-1 mutant (Fig. 3C,D). In contrast, the plasma membrane localization of PtdIns(4,5)P₂ was significantly decreased in the mss4-1 mutant (∼45.7%) and only slightly decreased in the stt4-1 pik1-1 double mutant (Fig. 3E,F). Changes in PtdIns(4)P and PtdIns(4,5)P₂ observed in stt4-1 pik1-1 double mutant at the restrictive temperature suggest that turnover of PtdIns(4,5)P₂ is slower than that of PtdIns(4)P, which is consistent with previous studies (Hammond et al., 2009, 2012).

To examine the effect each mutant has on endocytic internalization, we utilized three different endocytic markers. We first labeled wild-type and mutant cells with A594-α-factor and followed its localization at several time points after α-factor internalization. Similarly to the stt4 kd mutant, the stt4-1 mutant exhibited a slight delay in A594-α-factor internalization (Fig. 4A). The pik1-1 mutant showed little delay on its own, but displayed a severe delay when combined with the stt4-1 mutant (Fig. 4A), suggesting that the function of these proteins overlaps during endocytic internalization. These delays observed in stt4-1 and stt4-1 pik1-1 mutants were particularly pronounced in the mother cell (Fig. 4A). This might be because, in budding yeast, endocytosis is more pronounced in the bud than in the mother cell (Layton et al., 2011). Consistent with previous reports (Desrivières et al., 2002), the mss4-1 mutant also showed a severe endocytic defect (Fig. 4A). Quantitative analysis revealed that the mss4-1 mutant and the stt4-1 pik1-1 double mutant have obvious defects in A594-α-factor transport (Fig. 4B). We next examined the effect on endocytosis by assessing the internalization of ³⁵S-labeled α-factor. As shown in Fig. 4C, the stt4-1 mutant exhibited a defect in α-factor internalization, whereas the pik1-1 mutant had no apparent defect. Consistent with the analysis by using A594-α-factor, appreciable delays of ³⁵S-labeled α-factor internalization were observed in mss4-1 and stt4-1 pik1-1 double-mutant cells (Fig. 4C). We further examined the internalization of 3-triethylammoniumpropyl-4-p-diethylaminophenylhexatrienyl pyridinium dibromide (FM4-64), a lipophilic styryl dye that is used to follow bulk membrane. When added to wild-type cells, FM4-64 is immediately incorporated into the plasma membrane and internalized via bulk-phase endocytosis, and then transported to the vacuole within 20 min (Fig. 4D,E). Unexpectedly, this assay revealed that the stt4-1 pik1-1 mutant shows little delay in membrane internalization, whereas the mss4-1 mutant shows a remarkable delay (Fig. 4D,E).

To further investigate the effect of these mutants on clathrin coat assembly and membrane invagination, we examined the dynamics of the late coat module by using a GFP-tagged version of the actin cytoskeleton-regulatory complex protein Sla1p (Sla1-GFP) and those of actin patches by using mCherry-labeled actin-binding protein Abp1p (Abp1-mCherry) in the mss4-1 and stt4-1 pik1-1 mutants (Kaksonen et al., 2003). Interestingly, the mean lifetimes of Sla1-GFP and Abp1-mCherry patches were differentially affected by these mutants in a manner that was consistent with that seen for bulk lipids. In the stt4-1 pik1-1 mutant, Sla1-GFP patches had an average lifetime of 25.6±4.8 s, similar to that of Sla1-GFP patches in wild-type cells (27.5±5.0 s) (Fig. 5A,B). In contrast, the lifetime of Sla1-GFP patches was significantly increased in the mss4-1 mutant (49.3±17.6 s) (Fig. 5A,B). Live-cell imaging revealed that Sla1-GFP is internalized concomitantly with Abp1-mCherry under many conditions (Fig. 5A; Movie 1). And, similarly to Sla1p, the lifetime of Abp1p was considerably increased relative to wild type in the mss4-1 mutant (28.3±8.0 s), but not affected in the stt4-1 pik1-1 mutant (15.5±2.4 s) (Fig. 5C). Particle-tracking analysis revealed that, in wild-type and the stt4-1 pik1-1 cells, Abp1p patches first form at the plasma membrane and then move off the cortex (Fig. 5D, upper and middle panels; Movie 1), which is consistent with membrane invagination. However, in mss4-1 mutants, Abp1p patches form at the plasma membrane but do not move off the cortex before they disassemble (Fig. 5D, lower panel; Movie 1), indicating a failure in membrane invagination. Taken together with the results that the stt4-1 pik1-1 mutant showed decreased levels of PtdIns(4)P and that the mss4-1 mutant showed decreased levels of PtdIns(4,5)P₂ at the plasma membrane, these results suggest that PtdIns(4)P and PtdIns(4,5)P₂ are required for different steps in receptor-mediated endocytosis.

The epsin N-terminal homology (ENTH) and AP180 N-terminal homology (ANTH) domains have been identified as PtdIns-binding domains (Ford et al., 2001; Itoh et al., 2001). Therefore, we next examined if the decreased levels of PtdIns(4)P or PtdIns(4,5)P₂ affect the localization of ENTH/ANTH-domain proteins at the plasma membrane. Ent1p and Ent2p, are two essential proteins that belong to the epsin family, comprising an ENTH domain at their N-terminus (Aguilar et al., 2003; Chen et al., 1998; Wendland et al., 1999). Yap1801p, Yap1802p and Sla2p are homologs of the mammalian clathrin adaptor proteins CALM (also known as PICALM) and AP180 (Wendland and Emr, 1998), and Huntingtin-interacting protein 1-related protein (Hip1R), contain the ANTH domain. To precisely evaluate differences in the localization of these ENTH/ANTH-domain proteins, we compared each mutant directly to wild-type cells (Fig. 6A,B). Intriguingly, the fluorescence intensity of each type of protein patch was affected differently in stt4-1 pik1-1 and mss4-1 mutant strains (Fig. 6A,B). First, the fluorescence intensity of individual Sla2p patches was little affected in stt4-1 pik1-1 mutants (Fig. 6A,C), indicating that Sla2p localization does not depend on the level of PtdIns(4)P. In contrast, the localization of Ent1p, Ent2p, Yap1801p or Yap1802p was decreased to ∼75.2%, 63.7%, 51.8% or 21.6%, respectively, of that seen in wild type (Fig. 6A,C), indicating that recruitment of these proteins to endocytic sites is dependent on PtdIns(4)P. Consistent with the results that the plasma membrane localization of GFP-PH^Osh2 increases ∼2.4-fold in the mss4-1 mutant (Fig. 3D), we found that the fluorescence intensity of these proteins was increased ∼2.0-fold to 2.4-fold in the mss4-1 mutant (Fig. 6B,D). This observation supports the idea that localization of these ENTH/ANTH-domain proteins, except Sla2p, are dependent on the levels of PtdIns(4)P at the plasma membrane.

We have previously demonstrated that clathrin-coated pits (CCPs) first assemble at the plasma membrane, followed by recruitment of α-factor-bound Ste2p to CCPs (Toshima et al., 2006, 2009). Since the stt4-1 pik1-1 mutant showed a defect in α-factor internalization but did not show any defect in endocytic vesicle formation or bulk membrane internalization, we next examined ligand-induced receptor recruitment to CCPs by using the A594-α-factor (Toshima et al., 2009). To this end, we treated cells that expressed Sla1-GFP with latrunculin A (LatA), which leads to disassembly of cortical actin and blockage of the endocytic pathway at the internalization step (Kaksonen et al., 2003). Then, we incubated cells with A594-α-factor, and observed the localization of A594-α-factor at the plasma membrane. In wild-type cells treated with LatA, A594-α-factor was first observed as dispersed, homogeneous and faintly diffuse punctae when cells were first labeled (Fig. 7A; 0 min); it then concentrated into endocytic sites labeled by Sla1-GFP (∼75.2%) (Fig. 7A,B, 20 min) (Toshima et al., 2009). However in the stt4-1 pik1-1 mutant treated with LatA much less colocalization of A594-α-factor punctae with Sla1p patches was observed at 0 min (∼13.2%) or at 20 min (∼29.1%) (Fig. 7A bottom left, and B). A similar but relatively mild defect was also observed in the stt4 kd mutant following treatment with LatA (Fig. 7A top right, and B). Supporting the conclusion that these mutants have a specific defect in ligand-induced receptor recruitment to CCPs is the finding that the number of Sla1p patches was not significantly changed in stt4-1 pik1-1 and stt4 kd mutants when compared with wild type (Fig. 7C). By contrast, LatA treatment of the mss4-1 mutant resulted in remarkable colocalization of A594-α-factor punctae with Sla1p patches at 20 min (∼70%) (Figs 7A bottom and right panels, and 6B). We note that the number of Sla1p patches increased in the mss4-1 mutants (Fig. 7C), probably because of severe defects in endocytic internalization. These results clearly indicate that PtdIns(4)P, but not PtdIns(4,5)P₂, is required for Ste2 receptor recruitment to CCPs after binding α-factor.

Although roles for PtdIns(4,5)P₂ in endocytosis have been supported by a number of previous studies, what role PtdIns(4)P plays exactly was still unclear. In our study here, we have demonstrated that plasma membrane PtdIns(4)P is required for the efficient recruitment of ligand-bound Ste2p receptor to CCPs. We have also showed that the localization of the ENTH/ANTH-domain proteins, such as Ent1p, Ent2p, Yap1801p and Yap1802p, at endocytic sites was decreased in cells depleted of PtdIns(4)P. In a previous study, we had determined that ligand-induced recruitment of Ste2p to CCPs is mediated by phosphorylation and subsequent ubiquitylation of the C-terminal cytoplasmic tail of Ste2p (Toshima et al., 2009). While both modifications are essential for receptor recruitment to CCPs, ubiquitylation is the primary requirement because fusion of ubiquitin to the C-terminus of Ste2-6SA, an unphosphorylatable form of Ste2p, results in the efficient recruitment of the protein to CCPs (Toshima et al., 2009). Ent1p, Ent2p and Ede1p (the mammalian homolog of EPS15) contain ubiquitin-binding domains that directly interact with mono-ubiquitin and function in receptor internalization (Shih et al., 2002), and participate in recruitment of Ste2p receptor to CCPs (Toshima et al., 2009). Thus, insufficient localization of Ent1p and Ent2p at endocytic sites may decrease the efficiency of Ste2p recruitment to CCPs in stt4-1 pik1-1 mutant cells.

Based on the observation that the ENTH/ANTH domains preferentially bind to PtdIns(4,5)P₂, it was thought that PtdIns(4,5)P₂ might be a key regulator of clathrin coat assembly, which anchors endocytic coat proteins to the plasma membrane, (Ford et al., 2001; Itoh et al., 2001; Legendre-Guillemin et al., 2004). In contrast, PtdIns(4)P has been considered an immediate precursor of PtdIns(4,5)P₂ at the plasma membrane, although it plays essential roles in vesicle trafficking regulation at the Golgi complex (Wang et al., 2003). PtdIns(4)P is highly enriched in the Golgi complex, and many PtdIns(4)P effectors that localize at the Golgi − including GGA2 and epsinR − have been identified (Demmel et al., 2008; Hirst et al., 2003; Mills et al., 2003; Wang et al., 2007). EpsinR contains an ENTH domain that, essentially, has the same structure as that in epsin1 but that preferentially binds to PtdIns(4)P rather than PtdIns(4,5)P₂ (Mills et al., 2003). The ENTH domain of yeast Ent1p is similar to that of epsin1 and binds more strongly to PtdIns(4,5)P₂; however, this ENTH domain also has similarity to that of epsinR, which preferentially binds to PtdIns(4)P (Mills et al., 2003). Our data demonstrate that the localization of Ent1p, Ent2p, Yap1801p and Yap1802p is decreased in the PtdIns(4)P-depleted stt4-1 pik1-1 mutant, but increased in the PtdIns(4)P-accumulating mss4-1 mutant, suggesting that the ENTH/ANTH domains of these proteins can bind to PtdIns(4)P as well as PtdIns(4,5)P₂. Considering previous observations that ANTH and ENTH domains have the ability to bind PtdIns(4)P, although those interactions are usually less efficient than those with PtdIns(4,5)P₂ (Ford et al., 2001; Itoh et al., 2001), PtdIns(4)P might be required for the initial anchoring of endocytic coat proteins to the plasma membrane, and PtdIns(4,5)P₂ might strengthen the association of these proteins to clathrin-coated vesicles.

We observed distinct effects of the different lipid kinases on the recruitment of endocytic proteins acting in different steps in the process. Endocytic proteins have been divided into several modules (Kaksonen et al., 2005; Weinberg and Drubin, 2012), and ENTH/ANTH-domain proteins were categorized into coat modules, with Yap1801p and Yap1802p comprising the early coat module, and Ent1p, Ent2p and Sla2p comprising the mid coat module. We have demonstrated here that the localization of all these early and mid coat module proteins − except Sla2p − is reduced at the plasma membrane of stt4-1 pik1-1 mutants, which is consistent with the idea that Sla2p is differentially regulated than other ENTH/ANTH proteins. This is consistent with previous studies, showing that complete deletion of the ANTH domain in Sla2p does not impact its recruitment to endocytic sites (Sun et al., 2005). The localization and dynamics of Sla1p and Abp1p − which belong to the late coat protein and actin module, respectively − were also not affected in the stt4-1 pik1-1 mutant. Thus, the later steps of endocytosis, including Sla2p recruitment, seem to be preserved in PtdIns(4)P-depleted cells.

Although the recruitment of these ENTH/ANTH-domain proteins is strictly regulated, how this correlates to the timing of the appearance of PtdIns(4)P or PtdIns(4,5)P₂ at endocytic sites has not been clarified. Several studies have suggested that PtdIns(4,5)P₂ turnover is much slower than that of PtdIns(4)P, because depletion of cellular PtdIns(4)P causes only minor effects on the plasma membrane localization of PtdIns(4,5)P₂ (Hammond et al., 2009, 2012). Thus, a spatially restricted and rapid change from PtdIns(4)P to PtdIns(4,5)P₂ might not be required for endocytic coat assembly or internalization. This idea is supported by our observations that both the PtdIns(4)P probe GFP-PH^Osh2 and the PtdIns(4,5)P₂ probe GFP-PH^PLCδ do not display an endocytic patch-like localization. Taken together, these results suggest that a steady-state pool of PtdIns(4)P and PtdIns(4,5)P₂ is fundamentally required for receptor-mediated endocytosis, but that rapid changes in the levels of these lipids are not important for the initiation and progression of endocytosis.

Our observation that the dynamics of endocytic patches in the stt4-1 pik1-1 mutant are distinct from those seen in the mss4-1 mutant also clearly indicates distinct roles for PtdIns(4)P and PtdIns(4,5)P₂. Consistent with a previous study that used a different mss4^ts mutant (Sun and Drubin, 2012), Sla1p/Abp1p patches in our mss4-1 mutant form at the plasma membrane but do not move off the cortex before they assemble. This indicates that endocytic patches normally form but endocytic internalization is impaired in PtdIns(4,5)P₂-depleted cells. Abp1p patches persist considerably longer in the mss4-1 mutant than in the wild-type, suggesting that the primary role of PtdIns(4,5)P₂ during endocytosis is endocytic internalization mediated by actin assembly. In contrast, the lifetime and dynamics of Sla1p/Abp1p patches in the stt4-1 pik1-1 mutant are almost identical to those in wild-type cell. Several studies have indicated that PtdIns(4)P is not required to maintain the steady-state pool of PtdIns(4,5)P₂ at the plasma membrane (Hammond et al., 2009, 2012; Willars et al., 1998). Consistent with these observations, our results also show that the stt4-1 pik1-1 mutant significantly affects the synthesis of PtdIns(4)P but not of PtdIns(4,5)P₂ at the plasma membrane. Taken together, our results suggest that PtdIns(4)P and PtdIns(4,5)P₂ have different roles during receptor-mediated endocytosis, affecting coat protein-mediated receptor recruitment and plasma membrane invagination, respectively.

The yeast strains used in this study are listed in Table S1. All strains were grown in standard rich medium (YPD) or synthetic medium (SM) supplemented with 2% glucose and appropriate amino acids. C-terminal GFP or mCherry tagging of proteins was performed as described previously (Longtine et al., 1998). The stt4-1 mutant was generated as follows: The NotI-SacII fragment, which contains the S. cerevisiae ADH1 terminator and the URA3MX6 module, was amplified by PCR, and inserted into NotI- and SacII-digested pBluescript II SK (pBS-TADH-URA3). To create a plasmid library containing STT4 gene fragments (nt 4428-5700) carrying various mutations, error-prone PCR products amplified by JT2057 and JT2058 (Table S2), using yeast genome DNA as a template, were digested with BamHI and NotI and inserted into the BamHI and NotI-digested pBS-TADH-URA3. To integrate the plasmid library at the endogenous locus of the STT4 gene, the plasmid was linearized by NruI and transformed into wild-type cells; transformants were subsequently grown on SD-URA plates at 25°C. After 3−4 days, ∼1000 transformants were replica-plated on SD-URA plates and subsequently grown at 25 or 37°C for 2-3 days. The pik1-1 and mss4-1 mutants were generated by applying the same method, using primers listed in Table S2.

Fluorescence microscopy was performed using an Olympus IX81 microscope equipped with an ×100/NA 1.40 (Olympus) objective and an Orca-AG cooled CCD camera (Hamamatsu), using Metamorph software (Universal Imaging). Simultaneous imaging of red and green fluorescence was performed using an Olympus IX81 microscope as above, and an image splitter (Dual-View; Optical Insights) that divided the red and green components of the images with a 565-nm dichroic mirror and passed the red component through a 630/50 nm filter and the green component through a 530/30 nm filter. FM4-64 staining was performed as described previously (Toshima et al., 2005).

Fluorescence labeling of α-factor was performed as described previously (Toshima et al., 2006). For endocytosis assays, cells were grown to an OD₆₀₀ of ∼0.5 in 0.5 ml YPD, briefly centrifuged, and resuspended in 20 µl SM with 5 µM Alexa-Fluor-labeled α-factor. After incubation on ice for 2 h, the cells were washed with ice-cold SM. Internalization was initiated by addition of SM containing 4% glucose and amino acids at 25°C.

Preparation and internalization of ³⁵S-labeled α-factor was performed as described previously (Toshima et al., 2005). For the binding assay, cells were grown to an OD₆₀₀ of ∼0.3 in 1 ml YPD at 25°C, briefly centrifuged, and resuspended in 50 µl SM with 1% (w/v) BSA and ³⁵S-labeled α-factor on ice. After incubation on ice for 2 h, cells were washed with ice-cold SM and measured for their radioactivity.

Immunoblot analysis was performed as described previously (Toshima et al., 2005). Anti-HA antibody (Roche, clone no. 12CA5) was used at a dilution of 1:500 and rabbit anti-mouse IgG-horseradish peroxidase (Amersham Biosciences, Piscataway, NJ) at 1:1000 dilution was used as the secondary antibody. Chicken polyclonal antibody recognizing GFP (GeneTex, catalog no. GTX124117) was used at a dilution of 1:10,000 and HRP-conjugated rabbit polyclonal antibody recognizing chicken IgY (Promega, G135A) at 1:10,000 dilution was used as the secondary antibody. Anti-GAPDH antibody (Gene Tex, clone no. GT239) was used as a loading control. Immunoreactive protein bands were visualized by using the WesternLightning Plus ECL (PerkinElmer).

The fluorescence intensity of GFP-PH^Osh2 or GFP-PH^PLCδ in wild-type cells was calculated by subtracting average fluorescence intensity in the cytosol from that of a randomly selected area (1×1 pixel area, of at least 30 areas) in the plasma membrane or Golgi complex. For relative fluorescence intensity in mutants, the average fluorescence intensity in mutant cells was divided by the average fluorescence intensity in wild-type cells. Fluorescence intensity of cortical patch proteins in wild-type cells was calculated by averaging the fluorescence intensities of the center area (1×1 pixel area) of individual patches. For the relative fluorescence intensity in mutants, the average fluorescence intensity of the center area of the patch in mutant cells was divided by the average fluorescence intensity of this area in wild-type cells. The fluorescence intensities were analyzed by using the program ImageJ V1.44. Statistical tests were performed using GraphPad Prism version 6 (GraphPad Software, La Jolla, CA).

We thank Reina Shindo (Tokyo University of Science, Tokyo, Japan) for construction of plasmids and strains.