Epsin contains a phospholipid-binding ENTH domain coupled to C-terminal domain motifs that bind coated pit proteins. We examined how these domains interact to influence epsin function and localization in Dictyostelium. Although not required for global clathrin function, epsin was essential for constructing oval spores during development. Within the epsin protein, we found that features important for essential function were distinct from features targeting epsin to clathrin-coated pits. On its own, the phospholipid-binding ENTH domain could rescue the epsin-null phenotype. Although necessary and sufficient for function, the isolated ENTH domain was not targeted within clathrin-coated pits. The C-terminal domain containing the coated-pit motif was also insufficient, highlighting a requirement for both domains for targeting to coated pits. Replacement of the ENTH domain by an alternative membrane-binding domain resulted in epsin that sequestered clathrin and AP2 and ablated clathrin function, supporting a modulatory role for the ENTH domain. Within the ENTH domain, residues important for PtdIns(4,5)P2 binding were essential for both epsin localization and function, whereas residue T107 was essential for function but not coated pit localization. Our results support a model where the ENTH domain coordinates with the clathrin-binding C-terminal domain to allow a dynamic interaction of epsin with coated pits.
Clathrin-mediated endocytosis is a highly conserved process in which specific cargo on the plasma membrane is selected and internalized. Clathrin triskelia, key structural proteins of this process, are recruited to the membrane and assemble into coated pits that encompass endocytic cargo. These pits subsequently pinch off to form intracellular clathrin-coated vesicles. A wide variety of adaptors and accessory proteins select appropriate cargo and help recruit clathrin to the membrane. Epsin is one such clathrin adaptor.
First identified as a binding partner of epidermal growth factor receptor substrate 15 (Eps15) (Chen et al., 1998), epsin is thought to contribute to coated vesicle function in eukaryotic cells. Members of the epsin family share a similar structure. At the N-terminus, epsin contains an ENTH domain (epsin N-terminal homology) that binds specifically to the lipid PtdIns(4,5)P2 (Itoh et al., 2001). At the C-terminus, epsin contains several short binding motifs specific for clathrin and clathrin adaptors such as assembly protein AP2 and Eps15-homology (EH)-domain proteins (Chen et al., 1998; Kay et al., 1999). Accordingly, mammalian epsin coprecipitates with clathrin, AP2 and Eps15 in vitro (Chen et al., 1998; Owen et al., 1999; Traub et al., 1999) and colocalizes with clathrin and various endocytic adaptors in vivo (Chen et al., 1998; Drake et al., 2000; Newpher et al., 2005). Epsin promotes clathrin assembly (Ford et al., 2002; Kalthoff et al., 2002a) and is present, but not necessarily enriched, in purified clathrin-coated vesicles (Chen et al., 1998; Hawryluk et al., 2006). Epsins from different species also contain one or more ubiquitin-interacting motifs (UIMs) that interact with ubiquitylated cargo (Hofmann and Falquet, 2001; Polo et al., 2002; Aguilar et al., 2003; Barriere et al., 2006).
This modular organization suggests a model where the C-terminus of epsin acts as a scaffold for clathrin, clathrin adaptors and specific cargo, whereas the N-terminal ENTH domain tethers and promotes invagination of the coated pit from the plasma membrane. However, how these modules cooperate to facilitate epsin function in living cells remains unclear. Furthermore, domain analysis of fly and yeast epsin has led to the puzzling result that expression of solely the ENTH domain rescues phenotypic deficiencies in these organisms, suggesting that the C-terminus is dispensable (Wendland et al., 1999; Aguilar et al., 2003; Overstreet et al., 2003). The capacity of the isolated ENTH domain to function raises questions about what functional properties the C-terminal domain contributes to epsin.
Dictyostelium discoideum cells offer a model system where clathrin-coated pits associate with the plasma membrane, and clathrin is essential for important biological roles (O'Halloran and Anderson, 1992; Damer and O'Halloran, 2000; Wang et al., 2006). Moreover, Dictyostelium cells contain conserved adaptors that associate with clathrin on the plasma membrane (Stavrou and O'Halloran, 2006; Wang et al., 2006; Repass et al., 2007). In this study, we identified the Dictyostelium epsin ortholog epnA and found that it plays an essential role in spore development. In addition, our analysis highlighted separate and distinct contributions of the ENTH domain and the C-terminal domain to the localization and to the functional capacity of epsin. We conclude that the ENTH domain cooperates with the C-terminal domain of epsin to facilitate a dynamic interaction with clathrin-coated pits at the plasma membrane.
Identification of Dictyostelium epsin
By searching for genes that shared amino acid sequences similar to the ENTH domain of human Epsin-1 (EPN1), we identified the Dictyostelium discoideum ortholog of epsin from the Dictyostelium genome database (see Materials and Methods). We identified a single gene, which we named epnA, with high amino acid sequence identity (48%) to the Epsin-1 ENTH domain. This was the sole gene that contained an ENTH domain. From this we concluded that Dictyostelium contains a single gene for epsin. The predicted amino acid sequence for epnA contained multiple short binding motifs for other endocytic adaptors (Fig. 1A), consistent with epsin from other species (Salcini et al., 1997; Chen et al., 1998; Kay et al., 1999; Traub et al., 1999; Cadavid et al., 2000). In addition, Dictyostelium epsin also contained two Type I L(L,I)(D,E,N)(L,F)(D,E,S) clathrin-binding motifs (Dell'Angelica et al., 1998; Drake et al., 2000; ter Haar et al., 2000) (Fig. 1A). However, unlike most epsins in other species, the predicted amino acid sequence for Dictyostelium epsin did not contain a UIM (Hofmann and Falquet, 2001; Polo et al., 2002; Aguilar et al., 2003; Barriere et al., 2006). In this respect, Dictyostelium epsin was similar to Arabadopsis epsin, which also lacks a UIM (Holstein and Oliviusson, 2005). To confirm the ability of Dictyostelium epsin to bind to clathrin, we performed a pulldown binding assay. Bacterially expressed maltose-binding protein (MBP):epsin fusion protein was bound to amylose resin and incubated with Dictyostelium cell lysate. Analysis of the bound and unbound fractions revealed that clathrin sedimented with MBP:epsin, but not MBP alone (Fig. 1B). Under these conditions, we were not able to detect binding between epsin and AP2 (Fig. 1B).
To determine the cellular location of Dictyostelium epsin, we cloned a cDNA for epnA fused to GFP (green fluorescent protein) and expressed this epsin:GFP fusion construct in a wild-type background. As with mammalian epsins (Chen et al., 1998), Dictyostelium epsin showed a punctate distribution largely restricted to the plasma membrane, with some intracellular puncta (Fig. 1C). The epsin puncta colocalized with clathrin on the plasma membrane and also with intracellular clathrin puncta (Fig. 1D; Fig. 4C). Dictyostelium epsin puncta also colocalized extensively with AP2 at the plasma membrane (Fig. 1E; Fig. 4B). Thus both the domains and localization of Dictyostelium epsin are similar to epsins from other organisms.
Epsin-null mutants display limited clathrin-associated phenotypes and have abnormal spore morphology
To examine the contribution of Dictyostelium epsin to cellular functions, we used targeted gene replacement to generate two epsin-null mutants. The deletion of the epnA gene in these mutants was confirmed by PCR of genomic DNA (data not shown), and the absence of epsin protein expression was demonstrated by immunoblotting with anti-epsin antibodies (Fig. 2A). Subsequent experiments revealed no differences in phenotype between the two epsin-null cell lines.
Reconstitution experiments with purified proteins and liposomes suggest that epsin functions to invaginate clathrin-coated pits (Ford et al., 2002). If epsin contributes this essential role to clathrin-coated vesicle formation in living cells, epsin-null cells would be expected to exhibit clathrin-related phenotypic deficits. To test whether clathrin-mediated cellular functions were compromised by the loss of epsin, we assessed the epsin-null mutants for phenotypes displayed by clathrin mutants. These phenotypes include defects in osmoregulation in hypo-osmotic conditions, deficiencies in fluid-phase endocytosis, and abnormal development into fruiting bodies (O'Halloran and Anderson, 1992; Niswonger and O'Halloran, 1997a; Wang et al., 2003). All of these processes were normal in the epsin-null mutants (Fig. 2B and data not shown), suggesting that epsin is not critical for general clathrin function. Both clathrin heavy chain (CHC)-null and clathrin light chain (CLC)-null mutants are known to fail in cytokinesis when grown in suspension cultures (Niswonger and O'Halloran, 1997a; Wang et al., 2003). Similarly to the clathrin mutants, epsin-null cells also accumulated multiple nuclei when grown in suspension cultures (Fig. 2D,F). The absence of many phenotypes characteristic of clathrin mutants suggested that epsin does not supply an essential and global function, such as invagination, to clathrin-coated pit formation. Rather the discrete phenotype suggests that epsin contributes to a subset of clathrin function that includes cytokinesis.
In contrast to clathrin-null cells, epsin-null mutants developed normally into fruiting bodies (Fig. 2B). However, we noted an abnormal phenotype when examining the morphology of spores within mature fruiting bodies. Spores from wild-type fruiting bodies were oblong, but spores from epsin-null fruiting bodies were round (Fig. 2C,E). Measurement of the width:length ratio of wild-type spores was 0.60±0.01 (n=50; mean ± s.e.m.),whereas spores derived from epsin-null mutants had a width:length ratio of 0.88±0.02 (n=50) (Fig. 2D,F). This round spore phenotype was reminiscent of Dictyostelium Hip1r, another clathrin accessory protein (Repass et al., 2007). The restricted phenotype during development supported an essential role for epsin in a specialized pathway that controls the correct morphology of spores.
Clathrin and AP2 assemble into puncta on the membrane of epsin-null cells.
Epsins contain domains and motifs that bind plasma membrane lipids as well as clathrin and clathrin adaptors. We therefore tested whether epsin was essential for clathrin pit organization by assessing the ability of clathrin and the clathrin adaptor AP2 to assemble into puncta on the plasma membrane of epsin-null cells. Wild-type cells and epsin-null mutants were transformed with GFP-CLC, a marker known to reflect the endogenous distribution of clathrin (Wang et al., 2006), and then were immunostained with antibodies to AP2. In wild-type cells, clathrin formed puncta on the plasma membrane and in the cytoplasm (Fig. 3A). Clathrin puncta on the plasma membrane of wild-type cells colocalized extensively with AP2 (Fig. 3A, inset). In epsin-null cells, clathrin and AP2 puncta also formed, and the frequency and distribution of the two proteins were indistinguishable from that in wild-type cells (Fig. 3B). Subcellular fractionation of wild-type cells showed that clathrin partitioned into the low-speed and the high-speed membrane fractions. Clathrin showed a similar association with membrane fractions in epsin-null cells (Fig. 3C). Together, these observations suggested that epsin does not play an essential role in organizing clathrin or AP2 in coated pits.
Epsin localization into puncta on the plasma membrane requires clathrin
To address whether clathrin is required for the association of epsin with the plasma membrane, we examined the distribution of epsin tagged with GFP in clathrin-null and AP2-null mutants. Both the cytokinesis and spore morphology defects of epsin mutants were completely rescued by expression of epsin:GFP (Fig. 2C-F), demonstrating that epsin:GFP was functional, and that the deficiencies displayed by epsin-null cells were specific for the absence of epsin. In clathrin-heavy-chain mutants, clathrin-coated pits are absent (O'Halloran and Anderson, 1992). Likewise, epsin:GFP did not cluster into puncta in clathrin-heavy-chain mutants, but instead uniformly decorated the plasma membrane (Fig. 3D). In Dictyostelium cells that lack CLC, clathrin function is diminished, but the heavy chain remains assembled into puncta on the plasma membrane (Wang et al., 2003). In these CLC mutants, epsin:GFP distributed into puncta on the plasma membrane and colocalized with AP2 (Fig. 3F). These observations suggested that the CHC influences the distribution of epsin on the plasma membrane. Subcellular fractionation studies of epsin confirmed this influence. In wild-type cells, epsin fractionated with the membranes of the high-speed pellet. By contrast, epsin was found in the soluble high-speed supernatant in clathrin-heavy-chain mutants, indicating that clathrin-null mutants contained more soluble epsin than wild-type cells did (Fig. 3E).
Epsin does not require AP2 to associate with clathrin at the plasma membrane
The preceding experiments established that clathrin was an important determinant for the association of epsin with membranes and for clustering within puncta on the plasma membrane. In addition to motifs for binding clathrin, the C-terminus of epsin has motifs for binding AP2, the predominant and best characterized clathrin adaptor at the plasma membrane. To examine the contribution of AP2 to the cellular location of epsin, we expressed epsin:GFP in AP2α mutants lacking the large α subunit of AP2. Relative to wild-type cells, AP2α mutants show reduced numbers of clathrin puncta on the plasma membrane (our unpublished results). Nonetheless, epsin continued to colocalize with the remaining clathrin puncta in AP2α-null cells (Fig. 3G). Similarly to the reduced number of clathrin puncta on the membrane of AP2α mutants, epsin formed ∼20% fewer puncta on the plasma membrane of AP2α mutants (0.60±0.04 puncta/μm2, n=1047 puncta; 16 cells) compared with that in wild-type cells (0.77±0.04 puncta/μm2, n=683 puncta; 12 cells) (Fig. 3G). Subcellular fractionation of epsin in the AP2α mutants revealed that the association of epsin with membrane fractions was similar to that seen in wild-type cells (Fig. 3E). Taken together, these results indicate that, although AP2α is important for building clathrin-coated pits on the plasma membrane, AP2α is not a critical determinant for localizing epsin into coated pits.
The ENTH domain is required but is not sufficient for epsin association with clathrin and AP2 at the plasma membrane
In other organisms, the N-terminal ENTH domain of epsin has been shown to be sufficient for phenotypic rescue (Wendland et al., 1999; Aguilar et al., 2003; Overstreet et al., 2003). To explore the functional properties of Dictyostelium epsin in more detail, we generated two expression plasmids for GFP-tagged epsin truncations, epsin1-333 and epsin253-677, which separated the N-terminal ENTH domain from the C-terminal domain which contained motifs for binding clathrin, EH-domain proteins and AP2 (Fig. 4A).
We expressed these truncation constructs in an epsin-null background (Fig. 4D-E). Because the C-terminal domain of epsin contains motifs for binding AP2 and clathrin accessory proteins, we expected epsin253-677 to associate with the plasma membrane. However, examination by fluorescence microscopy of cells expressing epsin253-677 revealed that epsin253-677:GFP rarely localized to the plasma membrane, but instead associated with puncta in the cytoplasm (Fig. 4D,E compared with Fig. 4B,C). Thus the C-terminal domain of epsin associated with cytoplasmic puncta of clathrin and was excluded from plasma membrane clathrin puncta, contrary to the normal distribution for full-length epsin.
The complementary N-terminal epsin construct, epsin1-333, contained the complete ENTH domain plus a short, unstructured region. Consistent with the capacity of the ENTH domain to bind PtdIns(4,5)P2, epsin1-333:GFP localized uniformly on the plasma membrane and did not form discrete puncta (Fig. 4F,G). Epsin1-333:GFP also distributed along the plasma membrane of cells lacking CHC or CLC, as well as AP2α-null cells, suggesting that the ability of epsin1-333:GFP to associate with the plasma membrane was independent of clathrin or AP2 (supplementary material Fig. S1). Although the distribution of epsin1-333 was uniform, clathrin localized normally into puncta in both epsin-null and wild-type cells expressing epsin1-333:GFP (Fig. 4G and data not shown).
In addition to their localization, we also tested whether the N-terminal and the C-terminal truncations of epsin were able to rescue the phenotypic deficiencies of epsin-null cells. We tested the ability of both constructs to rescue cytokinesis by examining whether multinucleated cells accumulated in cultures of epsin mutants expressing either of the two constructs. Quantification of multinucleated cells in suspension cultures revealed that both epsin253-677:GFP and epsin1-333:GFP rescued the cytokinesis defect of epsin mutants (Fig. 5C).
We also examined the ability of the two constructs to rescue the spore morphology defect of epsin mutants. Epsin-null cells expressing epsin253-677:GFP developed into fruiting bodies containing round spores that were indistinguishable from the control epsin-null mutants (Fig. 5A,B). The failure of epsin253-677:GFP to restore normal spore morphology was not an artifact of the GFP tag, because epsin-null cells expressing epsin253-677 without GFP also developed into fruiting bodies that contained round spores (Fig. 5B). By contrast, epsin-null mutants expressing the N-terminal construct epsin1-333:GFP developed into fruiting bodies that contained oblong spores indistinguishable from that in the wild type (Fig. 5A,B). Thus epsin253-677:GFP was able to rescue the cytokinesis failure but not the spore morphology defect, whereas epsin1-333 was able to fully rescue both phenotypic defects of epsin-null mutants.
A canonical PtdIns(4,5)P2-binding domain cannot substitute for the ENTH domain
The analysis of epsin domains suggested that the ENTH domain was both necessary and sufficient to target epsin to the membrane and to rescue the spore morphology and cytokinesis defects of epsin-null mutants. A significant function of the epsin ENTH domain is to bind PtdIns(4,5)P2 (Itoh et al., 2001; Ford et al., 2002). We therefore asked whether another PtdIns(4,5)P2-binding domain could functionally replace the ENTH domain of Dictyostelium epsin. The PH domain of mammalian PLCδ, a canonical PtdIns(4,5)P2-binding domain, is of comparable size to the ENTH domain and also binds to PtdIns(4,5)P2 (Lemmon et al., 1995; Stauffer et al., 1998; Itoh et al., 2001). We generated a construct that tagged the PLCδPH domain with GFP and examined its distribution in epsin-null and wild-type cells. Consistent with membrane-binding properties similar to the ENTH domain of epsin, PLCδPH:GFP displayed a uniform plasma membrane localization comparable with epsin1-333:GFP and did not disrupt clathrin localization or function (supplementary material Fig. S2).
To determine whether this canonical PtdIns(4,5)P2-binding domain could substitute for the ENTH domain function, we made a chimeric GFP-tagged epsin that replaced the ENTH domain with the PH domain of PLCδ (Fig. 4A). If this alternative PtdIns(4,5)P2-binding domain was able to substitute for the ENTH domain, the PH-epsin C-terminal domain chimera (PH:epsin253-677:GFP) should distribute similarly to full-length epsin. However, when expressed in wild-type cells, the PH:epsin253-677:GFP chimera localized in a distinct and aberrant pattern. Instead of forming puncta evenly distributed on the plasma membrane and puncta within the cytoplasm, PH:epsin253-677:GFP aggregated into large patches on the plasma membrane (Fig. 6A,B compare with Fig. 4B,C). Moreover, these aberrant patches sequestered both AP2 and clathrin. Staining with anti-AP2 antibody revealed that AP2 puncta frequently clustered within the PH:epsin253-677:GFP patches (Fig. 6A). Staining with anti-clathrin antibody revealed that PH:epsin253-677:GFP caused severe mislocalization of clathrin. In wild-type cells or epsin-null cells rescued with epsin:GFP, clathrin normally forms discrete puncta on the plasma membrane and in the perinuclear region of the cytoplasm (Fig. 4C). However, in wild-type cells expressing PH:epsin253-677:GFP, clathrin aggregated together with PH:epsin253-677:GFP caps at the plasma membrane, and nearly all cytoplasmic and perinuclear clathrin staining was absent (Fig. 6B). Epsin-null cells expressing PH:epsin253-677:GFP showed an identical distribution and mislocalization of AP2 and clathrin (data not shown). Thus, substitution of an alternative PH domain for the ENTH domain allowed the chimeric epsin to associate with the plasma membrane, and allowed the C-terminal domain to bind clathrin and AP2. However, the chimeric epsin also sequestered AP2 and clathrin into abnormal patches on the plasma membrane.
Imaging living cells expressing the chimeric epsin revealed that its dynamic association with the plasma membrane was also aberrant (Fig. 6C). Puncta of GFP-epsin associated transiently with the plasma membrane and GFP-epsin could be seen to build up into a discrete spot that subsequently disappeared. By contrast, the patches of PH:epsin253-677:GFP were relatively more static on the membrane and did not appear to form or disassemble. Both the PH:epsin253-677:GFP chimera and epsin GFP were expressed in similar amounts (data not shown). We therefore concluded that substitution of the PH domain for the ENTH domain disrupted the capacity of epsin to form transient puncta on the plasma membrane.
Expression of PH:epsin253-677:GFP impairs clathrin function
To determine whether the sequestration of clathrin on the membrane by PH:epsin253-677 ablated clathrin function, we tested wild-type cells expressing PH:epsin253-677:GFP for phenotypes typical of clathrin mutants: defective osmoregulation, cytokinesis failure and abnormal development.
Clathrin mutants display defects in the size and activity of the contractile vacuole, an osmoregulatory organelle in Dictyostelium, and are therefore osmosensitive (O'Halloran and Anderson, 1992; Wang et al., 2003). To test whether expression of PH:epsin253-677:GFP induced osmosensitivity, we shifted cells from medium to water, and examined the contractile vacuole under differential interference contrast (DIC) microscopy. Wild-type cells displayed an increase in contractile vacuole activity, with the contractile vacuole swelling and discharging. Similarly to clathrin-light-chain mutants (Wang et al., 2003), wild-type cells expressing PH:epsin253-677:GFP developed abnormally large contractile vacuoles with prolonged cycles of expansion (Fig. 6D). This effect was due to the expression of PH:epsin253-677:GFP because the contractile vacuole activity was not altered in wild-type cells expressing full-length epsin, either of the two epsin truncations or the PH domain alone (data not shown).
Clathrin is also critical for cytokinesis. Both CLC and CHC mutants are unable to divide in suspension cultures and become large and multinucleated (Niswonger and O'Halloran, 1997a; Wang et al., 2003). When grown in suspension cultures, wild-type cells expressing PH:epsin253-677:GFP also accumulated many nuclei to the same extent as CHC mutant cells, demonstrating a similarly severe defect in cytokinesis (Fig. 5C; Fig. 7E). These defects showed that coupling the C-terminus of epsin to an alternative membrane-binding domain induces dominant-negative phenotypes in growing cells that are characteristic of clathrin mutants.
In contrast to the other clathrin deficiencies induced by expressing PH:epsin253-677:GFP in wild-type cells, expression of PH:epsin253-677:GFP did not impair wild-type cells during development. Clathrin mutants are not able to complete development and aggregate to form stunted structures (Niswonger and O'Halloran, 1997b; Wang et al., 2003). However, wild-type cells expressing PH:epsin253-677:GFP developed into fruiting bodies with a stalk and a sorus that appeared normal in structure (data not shown). Moreover, the fruiting bodies of wild-type cells expressing PH:epsin253-677:GFP contained oblong spores identical in shape to wild-type spores, indicating that PH:epsin253-677:GFP did not induce the formation of abnormal spores (Fig. 5A,B). Nonetheless, whereas the chimeric PH:epsin253-677:GFP protein did not lead to dominant-negative developmental phenotypes, the chimeric epsin also did not rescue the spore defect of epsin-null cells. Examination of the spores housed within the sori of epsin-null cells expressing the chimeric PH:epsin253-677:GFP revealed round spores that were identical in morphology to the spores of epsin mutants (Fig. 5A,B).
Identification of residues in the ENTH domain important for localization and function
The inability of the PH domain to substitute for the ENTH domain suggested that the ENTH domain contributed more than PtdIns(4,5)P2-binding activity to epsin. To determine how the ENTH domain contributed to epsin localization and function, we first asked whether the PtdIns(4,5)P2-binding ability of the ENTH domain was critical for epsin function. Amino acids R65 and K78 have been shown to be critical for the interaction between the ENTH domain and PtdIns(4,5)P2 (Itoh et al., 2001). To directly test the importance of this activity, we constructed two plasmids to express mutant versions of either the ENTH domain or full-length epsin with the R65A/K78A mutations. Assessment of the PtdIns(4,5)P2-binding capacity of ENTHR65A/K78A confirmed that mutating these residues impaired the ability of the ENTH domain to bind PtdIns(4,5)P2 (Fig. 7A). Examination of the cells expressing the GFP-tagged proteins revealed that ENTHR65A/K78A and epsinR65A/K78A failed to associate with the plasma membrane (Fig. 7B,C), consistent with the insufficiency of the C-terminal domain to associate with clathrin-coated pits. Moreover, ENTHR65A/K78A and epsinR65A/K78A also failed to rescue the round spore phenotype, demonstrating that the ability to bind PtdIns(4,5)P2 was required for both epsin function and localization (Fig. 7D). We also tested the contribution of another amino acid within the ENTH domain, T107, a residue not predicted to function in PtdIns(4,5)P2 binding. The analogous amino acid in yeast epsin is essential for viability (Aguilar et al., 2006), demonstrating an important contribution to epsin activity; however, the contribution of this residue to epsin localization has not been examined. We therefore constructed a plasmid that introduced the T107A mutation into the ENTH domain and full-length epsin. In contrast to epsinR65A/K78A, the T107A mutation did not affect the localization of epsin. ENTHT107A was distributed uniformly along the plasma membrane and epsinT107A localized within puncta on the plasma membrane (Fig. 7B,C). Similarly to wild-type epsin, these puncta colocalized with plasma membrane clathrin (data not shown). However, despite the wild-type association with the plasma membrane or coated pits, neither ENTHT107A nor epsinT107A were able to rescue the round-spore phenotype of epsin-null cells (Fig. 7D). Thus this mutation separates the contribution of the ENTH domain to epsin localization from its contribution to essential cellular function.
Epsin is a phylogenetically conserved clathrin adaptor protein. Our results define determinants essential for targeting epsin into clathrin-coated pits that are distinct from determinants essential for epsin function. In this work, we identified clathrin, but not AP2, as essential for epsin localization within clathrin-coated pits. Our analysis also demonstrated that a cooperative interaction between the two domains of epsin enables this protein to interact dynamically with clathrin pits on the plasma membrane. Our results support a model where the ENTH domain coordinates with the clathrin-binding C-terminal domain to tether epsin to the plasma membrane for a productive and functional interaction with clathrin-coated pits. Independent of targeting to coated pits, the isolated ENTH domain is both necessary and sufficient for rescuing the aberrant spore morphology of epsin-null cells. Thus, determinants for targeting epsin to coated pits are distinct from those that supply function.
Clathrin, but not AP2, is a determinant for localizing epsin within coated pits
By examining the localization of epsin in different mutant backgrounds, we were able to define how other proteins contribute to the localization of epsin within clathrin-coated pits. In wild-type cells, double-label fluorescence microscopy revealed extensive colocalization between clathrin and epsin. A small number of epsin punctae were found without clathrin, but these static experiments could not distinguish whether clathrin was required for a particular phase of a dynamic coated pit or whether a small subset of epsin spots did not require clathrin. We addressed these two possibilities using clathrin mutants, and identified the CHC as necessary for epsin to cluster to plasma membrane puncta. Clathrin-heavy-chain-null mutants distributed epsin uniformly on the plasma membrane, as revealed by microscopy, and contained more soluble epsin than wild-type cells, as revealed by subcellular fractionation (Fig. 3D,E). Among Dictyostelium clathrin-associated proteins, this requirement for clathrin to form punctae is unique, because clathrin mutants continue to form puncta of AP180 and AP2 on their membranes (Stavrou and O'Halloran, 2006) (our unpublished results).
By contrast, epsin continued to cluster within clathrin-coated pits in AP2α mutants. Deletion of AP2 in Dictyostelium caused a marked decrease in the total number of puncta at the membrane that contain epsin and clathrin (our unpublished results), which is consistent with depletion experiments in vertebrate cell culture (Hinrichsen et al., 2003; Motley et al., 2003). Although the number of epsin puncta is reduced in AP2α mutants, the remaining epsin puncta continue to colocalize with clathrin. Thus, although AP2 increases the number of clathrin puncta on the membrane, the presence of AP2 is not critical for epsin to incorporate into clathrin-coated pits. This is in agreement with our biochemical results, which did not detect a physical interaction between epsin and AP2. Similarly, we have found that deletion of other Dictyostelium clathrin accessory proteins, including Hip1r and AP180, does not affect epsin localization to clathrin-coated pits (Stavrou and O'Halloran, 2006) (our unpublished results).
Dictyostelium contains a single epsin gene and did not appear to contain other homologs, including the related epsinR protein, which binds AP1, and is associated with Golgi trafficking events (Kalthoff et al., 2002b; Wasiak et al., 2002; Hirst et al., 2003; Mills et al., 2003). Despite the absence of related genes that could supply redundant function, we found that Dictyostelium epsin-null cells manifested only limited clathrin-associated phenotypes. In epsin-null cells, epsin function may be covered by other clathrin adaptors, including AP2 and AP180. Similarly to epsin, AP2 and AP180 bind PtdIns(4,5)P2 and contain multiple motifs for interacting with other coated pit proteins (Legendre-Guillemin et al., 2004; Edeling et al., 2006). Our findings in epsin-null mutants might indicate that epsin is not essential for initiating clathrin pit assembly, and are consistent with more specialized roles for epsin rather than assembly of the clathrin lattice itself.
Contribution of the C-terminal domain
In addition to examining how other proteins contribute to epsin localization, we also defined determinants within the epsin protein necessary and sufficient for targeting within coated pits. As with other epsins, the Dictyostelium epsin C-terminal domain contained motifs for interacting with coated pits. These motifs included clathrin- and AP2-binding motifs, but not a motif for interacting with ubiquitin. The latter motif is also lacking in Arabidopsis epsin (Holstein and Oliviusson, 2005). Surprisingly, we found that, although essential, the clathrin-binding C-terminal domain was not sufficient for associating with clathrin pits.
Coupling the C-terminal domain to an alternative membrane-binding domain, a PH domain, created a chimeric molecule capable of associating with clathrin and AP2 on the plasma membrane. Neither the C-terminal domain nor the PH domain on its own associated with clathrin on the membrane. Thus, the C-terminal domain contributes a potent capacity to associate with clathrin-coated pits, but only when targeted to the plasma membrane by a membrane-binding domain.
However, the chimeric PH:epsin253-677 molecule was not functional, and even sequestered clathrin to the extent of abolishing clathrin function. Thus the C-terminal domain is necessary but not sufficient for the functional interaction of epsin with clathrin-coated pits; the ENTH domain is also required. Moreover, the nonproductive and relatively static interaction of the chimeric molecule with clathrin at the plasma membrane suggests that the ENTH domain tempers the clathrin-binding ability of the C-terminal domain, allowing the interaction between epsin and clathrin to be transient, dynamic, and functional.
The ENTH domain is essential for localization and sufficient for function
The insufficiencies of the isolated C-terminal domain highlight a unique contribution of the ENTH domain to both the localization and function of epsin. The ENTH domain binds to PtdIns(4,5)P2 on the plasma membrane, a phospholipid critical for coated pit assembly (Zoncu et al., 2007). The C-terminal domain of epsin required this membrane-binding function of the ENTH domain in order to cluster within clathrin pits. However, the disruption of clathrin distribution and dominant-negative phenotypes associated with expression of the PH:epsin253-677 chimera suggested a new function for the ENTH domain in modulating the clathrin binding capacity of epsin.
Although the ENTH domain is essential for epsin localization, it is not sufficient. The ENTH domain alone could not cluster to clathrin-coated pits, but instead distributed uniformly over the plasma membrane. By contrast, the ENTH domain was both necessary and sufficient to rescue the spore morphology defects of epsin-null mutants. Thus, the determinants sufficient for clustering epsin within clathrin-coated pits, which require both the C-terminal domain and the ENTH domain, are distinct from the determinants sufficient for supplying function, which are contained solely in the ENTH domain.
Determinants within the ENTH domain that contribute to epsin function
Mutating residues R65 and K78 in the isolated ENTH domain ablated PtdIns(4,5)P2 binding and also ablated the capacity of the ENTH domain to rescue epsin-null phenotypes. Similarly, mutating R65/K78 in full-length epsin also rendered the protein unable to bind to coated pits and compromised its ability to rescue the round spore phenotype of epsin mutants. This mutant demonstrates that the ability of the ENTH domain to bind PtdIns(4,5)P2 is required for both epsin localization and essential function. By contrast, mutating the T107 residue rendered epsin non-functional, but still able to localize within clathrin-coated pits. Thus this residue within the ENTH domain contributes to the essential function of epsin, but does not contribute to the ability of epsin to target to and incorporate within clathrin-coated pits.
How does the T107 residue contribute to epsin function? The analogous residue in the yeast epsin homolog is part of a functional patch that binds to a GTPase-activating protein (GAP) for cell division control protein cdc42 and contributes to regulation of the actin cytoskeleton and cell polarity (Aguilar et al., 2006). The mechanism by which the T107 residue contributes to epsin function may be different than in yeast, because the Dictyostelium genome does not contain a gene for cdc42, although it does contain multiple genes that encode Rac GTPases. The Dictyostelium ENTH domain could supply a similar function by determining the polar organization of cellular components in the oblong spore. At present, little is known about how Dictyostelium spores construct their oblong shape. Interestingly, the clathrin accessory protein Hip1r also forms abnormally round spores in Dictyostelium (Repass et al., 2007). Both epsin-null cells and Hip1r-null cells produce round spores with slightly reduced viability, but Hip1r-null spores are more sensitive to heat and detergent treatment, indicating that the spore phenotype of Hip1r-null cells is more severe (Repass et al., 2007). Moreover, the ENTH domain of epsin is required for the phosphorylation and coated pit localization of Hip1r. An important function of the ENTH domain may be to regulate the localization and activity of Hip1r.
Functional contribution of the ENTH domain
The ability of the isolated ENTH domain to rescue epsin-null phenotypes even though it does not localize within coated pits suggests that epsin could have two distinct functions. One function, governed by the ENTH domain, is an essential developmental function that contributes to spore morphology. The capacity of the ENTH domain to function independently of clathrin-coated pits may be universal, as indicated by the ability of the isolated ENTH domains for yeast and Drosophila to rescue null phenotypes. Epsin does not require coated-pit localization in order to operate in this capacity, suggesting that this activity might be independent of clathrin. The other function of epsin is within clathrin-coated pits on the plasma membrane.
What is the functional contribution of epsin to clathrin-coated pits? In vitro studies demonstrate that epsin promotes invagination of clathrin assembled on lipid monolayers (Ford et al., 2002). More recently, in vivo studies have suggested that clathrin itself is the driving force in coated-pit invagination from the plasma membrane (Hinrichsen et al., 2006). Consistent with this observation, we found that Dictyostelium epsin-null cells manifested only limited clathrin-associated deficits. Our results argue for a more specialized role for epsin during clathrin-mediated endocytosis. Similarly, the Drosophila epsin homolog is not required for general clathrin-mediated endocytosis, but is specifically required for the appropriate processing of the Delta ligand during Notch signaling events (Overstreet et al., 2004; Wang and Struhl, 2004; Wang and Struhl, 2005).
It has been suggested that adaptors such as epsin are involved in sorting ligands to distinct endosomal populations (Lakadamyali et al., 2006) and allow precise endocytosis of certain surface receptors that are critical for appropriate cell fate specification (Berdnik et al., 2002; Overstreet et al., 2003; Traub, 2003; Wang and Struhl, 2005). Dissecting how epsin and other adaptors function in eukaryotic cells to tailor clathrin-coated pits amidst large volumes of membrane traffic remains an important challenge.
Materials and Methods
Strains and cell culture
Dictyostelium discoideum strains included Ax2, an axenic wild-type strain, 10G10 and 5B4, epsin-null strains derived from Ax2 (described below), 6A5, an α-adaptin-null line derived from Ax2, 5E2, a CHC-null line derived from Ax2 (Niswonger and O'Halloran, 1997b) and 2A1, a CLC-null strain derived from NC4A2 (Wang et al., 2003). Cells were cultured on tissue culture plates with HL-5 medium (Sussman, 1987) supplemented with 60 U/ml penicillin and 60 μg/ml streptomycin (Invitrogen, Carlsbad, CA) at 18°C. Null cells grown under selection were supplemented with 5 μg/ml blasticidin (ICN Biomedicals, Irvine, CA) and cells carrying expression plasmids were supplemented with 20 μg/ml G418 (geneticin, Gibco-BRL, Invitrogen).
Targeted replacement of epnA in Dictyostelium discoideum
Genomic sequence upstream of epnA was PCR-amplified using 5′-TTAAAAAAGGTAAAGATGCAGTATTG-3′ and 5′-TTGGAAATTTGGTGTTGCTGGTG-3′. Downstream genomic sequence was PCR-amplified using 5′-AATCAAAGTGGTGCGAATAGAAATAC-3′ and 5′-AATGATGATAGTAAAACTGATGGTAGAAG-3′. Genomic sequences were cloned on either side of the Bsr cassette in pSP27-BSR (Wang et al., 2002) using XhoI/HindIII (5′) and EcoRI (3′), generating the plasmid pSP72-BSR-EpsinKO. 10 μg linearized vector was transformed into Ax2 cells by electroporation. Cells were diluted into 96-well plates and grown under Bsr selection. Clonal transformants lacking the entire epnA gene were identified by western blot and PCR analysis, and two of these clones, 10G10 and 5B4, were selected for further study.
cDNA cloning and sequence analysis
The protein sequence of human Epsin-1 (EPN1; accession number NP037465) was used to search the Dictyostelium genome database (http://www.dictybase.org) for the best match using BLAST. A single homologous gene product was identified (DDB0183945, accession number XM630177). A complete cDNA clone was obtained from a Dictyostelium discoideum cDNA library using polymerase chain reaction (PCR) with primers 5′-TGGAGACTATGATTAAAAGTTATATTAAAAAAGGTAAAGATGCAGTATTGAATACACCAGAAATTGAAAGAAAGGTTAG-3′ and 5′-GCAGATCCCATGCTATTAGTATTTCTATTCGC-3′ and cloned into pCR2.1 using TA Cloning Kit (Invitrogen) to generate pCR2.1-Epsin. DNA sequences were managed using EditSeq and SeqMan (DNAStar, Madison, WI).
Cloning of expression plasmids
Epsin was cloned into pTX-GFP (Levi et al., 2000) (a kind gift from T. Egelhoff) to generate pTx-EpsinGFP using KpnI and EcoRV. The cDNA encoding the epsin1-333 truncation was PCR-amplified from pCR2.1-Epsin with 5′-TGGAGACTATGATTAAAAGTTATATTAAAAAAGGTAAAGATGCAGTATTGAATACA CCAGAAATTGAAAGAAAGGTTAG-3′ and 5′-GGTCGACTTCTTCCGCCAG-3′ and ligated into pCR2.1 to generate pCR2.1-epsin1-333. pCR2.1-epsin1-333 was then digested with EcoRI, blunt ended and cloned into the EcoRV site of pTxGFP, making pTX-pCR2.1-epsin1-333GFP. Epsin253-677 was amplified by PCR from pCR2.1-Epsin using 5′-TATAGTAATAGAGCAGGTGAGGAAACAAGAAG-3′ and 5′-CAGATCCCATGCTATTAGTATTTCTATTCGC-3′ and ligated into pCR2.1 to make pCR2.1-epsin253-677. Epsin253-677 was cloned from pCR2.1-epsin253-677. into pTx-GFP using BamHI and XhoI, making the expression plasmid pTx-GFP-epsin253-677. A cDNA encoding the PH domain of PLCδ (kind gift from Tobias Meyer, Stanford University, Stanford, CA) was cloned into pTxGFP using BamHI and NotI to generate pTx-GFP-PH. PH:epsin253-677 was generated by cloning the PH domain into pCR2.1-epsin253-677 with HindIII and SacI to make pCR2.1-PH:epsin253-677. PH:epsin253-677 was then cloned into pUC18 (Invitrogen) with HindIII and HincII to make pUC18-PH:epsin253-677. Finally, PH:epsin253-677 was cloned into pTx-GFP with KpnI, making pTX-PH:epsin253-677GFP. To generate pTX-epsin253-677, an expression plasmid for epsin253-677 without the GFP tag, we amplified a cDNA encoding epsin253-677 with 5′-CAGTGTGCTGGTACCCGGCTTTATAGTAATAG-3′ and 5′-GATGGATAGGATCCTAATTCGGCTTCAG-3′, ligated into pCR2.1, and cloned into pTxGFP with KpnI and BamHI, effectively replacing GFP with epsin253-677. Plasmid maps were managed using Gene Construction Kit (Textco BioSoftware, West Lebanon, NH).
Dictyostelium cell lines were transformed with various expression plasmids by electroporation. 5×106 cells in 100 μl buffer H-50 (20 mM HEPES, 50 mM KCl, 10 mM NaCl, 1 mM MgSO4, 5 mM NaHCO3, 1 mM NaH2PO4) were mixed with 10 μg plasmid and electroporated using a Bio-Rad Gene Pulser (Bio-Rad, Hercules, CA) at 75 kV and 25 μF.
Cells expressing GFP expression plasmids were harvested and allowed to attach to glass coverslips for 10 minutes at 18°C and incubated with low-fluorescence media (Liu et al., 2002) for at least 20 minutes. Cells were fixed with 2% formaldehyde and 0.01% Triton X-100 in PDF (2 mM KCl, 11 mM K2HPO4, 13.2 mM KH2PO4, 0.1 mM CaCl2, 0.25 mM MgSO4, pH 6.7) at room temperature for 15 minutes and then in 100% methanol at -20°C for 5 minutes, then rinsed with PDF and mounted on glass slides. For immunostaining, cells on coverslips were blocked with 3% BSA (Fisher Scientific, Pittsburgh, PA) in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, pH 7.4) for 20 minutes at 37°C, and then incubated with rabbit anti-CLC (Wang et al., 2003) or rabbit anti-AP2 IgG for 1 hour. Coverslips were rinsed with PBS, incubated for 30 minutes with 30 μg/ml goat anti-rabbit IgG conjugated to Texas Red (Molecular Probes, Invitrogen), and rinsed again in PBS. Coverslips were then rinsed in sterile, distilled water and mounted on glass slides. Images were taken using an inverted Nikon Eclipse TE200 microscope (Nikon Instruments, Dallas, TX) with 100× 1.4 NA PlanFluor objective and a Quantix 57 camera (Roper Scientific, AZ) controlled by Metamorph software (Universal Images, PA). Confocal images were acquired on a Leica TCS-SP2 laser-scanning confocal inverted microscope (Leica Microsystems, Wetzlar, Germany). Images were processed using Metamorph (Molecular Devices, Sunnyvale, CA) and Adobe Photoshop (Adobe Systems, San Jose, CA) software.
Live cell imaging by TIRF microscopy
Cells were imaged by TIRF (total internal reflection fluorescence) microscopy at Washington University School of Medicine, St Louis, Missouri. Cells were harvested, rinsed with PDF, and allowed to attach to acid-washed glass coverslips. Cells were imaged live on an Olympus 1X81 inverted microscope (Olympus America, Center Valley, PA) with an XR/Mega-10 camera (Stanford Photonics, Palo Alto, CA) using a 488 nm laser.
Generation of anti-epsin antibodies
A cDNA for epsin253-677 was cloned from pCR2.1-epsin253-677 into pMAL-C2X (New England Biolabs, Ipswich, MA) with BamHI and PstI so that epsin253-677 was downstream of maltose-binding protein (MBP), resulting in expression of MBP-epsin253-677 fusion protein from the plasmid pMAL-MBP-epsin253-677. The pMAL-MBP-epsin253-677 expression plasmid was transformed into Eschericha coli BL21 and the fusion protein was purified according to the manufacturer's protocol. Purified MBP-epsin253-677 was used to generate anti-epsin polyclonal antisera in rabbits (Cocalico Biologicals, Reamstown, PA).
Cells were harvested and allowed to attach to glass coverslips for 10 minutes at 18°C. Cells were then shifted to sterile, distilled water and imaged on an inverted Nikon Eclipse TE200 microscope using DIC optics.
Development and spores
To develop fruiting bodies, approximately 5×107 cells were harvested, washed with PDF and plated on starvation agar plates [0.2 mM CaCl2, 2.0 mM MgSO4, 20 mM MES pH 6.7, 1% Agar Noble (BD, Sparks, MD)]. Fruiting bodies were allowed to develop for 48 hours and then imaged using a Zeiss STEMI SR stereoscope (Carl Zeiss, Thornwood, NY). Spores were harvested from development plates by sharply striking the inverted plate on a hard surface and resuspending spores from the lid in PDF. Spores were plated on glass coverslips, allowed to settle, and imaged using an inverted Nikon Eclipse TE200 microscope with DIC optics.
Cytokinesis and growth in suspension
Cells were diluted to 1×104-5 cells/ml and grown in HL-5 on a rotary shaker at 218 r.p.m. at 18°C. Cultures were sampled periodically and counted on a hemocytometer. For DAPI (4′,6-diamidino-2-phenylindole) staining, cells were grown in suspension for 72 hours and then allowed to attach to glass coverslips for 10 minutes. Cells were fixed with 100% methanol at -20°C for 5 minutes and rinsed with PDF. Cells were then stained with 0.05 μg/ml DAPI (Invitrogen) in PDF for 10 minutes, rinsed, mounted on glass slides and imaged on an inverted Nikon Eclipse TE200 microscope.
Cells were fractionated according to Wang et al. (Wang et al., 2003). Briefly, cells were collected, washed, and resuspended to 4×107 cells/ml in MES isolation buffer (10 mM MES (pH 6.5), 50 mM potassium acetate, 0.5 mM MgCl2, 1 mM EGTA, 1 mM DTT, and 0.02% NaN3) with protease inhibitors (Fungal Protease Inhibitor cocktail, Sigma-Aldrich, St Louis, MO). Cells were lysed by passing through two pieces of Osmonics (GE Osmonics, Trevose, PA) polycarbonate membrane (pore size: 5 μm) in a Gelman Luer-Lock-style filter (Gelman Sciences, Ann Arbor, MI). Cell lysates were centrifuged at 3000 g for 10 minutes at 4°C to generate a low speed pellet (LSP) and low speed supernatant (LSS). The LSS was ultracentrifuged at 100,000 g for 60 minutes at 4°C, resulting in a high-speed supernatant (HSS) and a high-speed pellet (HSP).
Epsin/ENTHR65A/K78A and Epsin/ENTHT107A were generated using the Stratagene Quick Change Site-Directed Mutagensis Kit (Stratagene, La Jolla, CA) with primer pairs 5′-AATTATTATGGGTGTAATTTGGAAAGCTATTAATGATCCAGGCAAGTTTTGG-3′ and 5′-CCAAAACTTGCCTGGATCATTAATAGCTTTCCAAATTACACCCATAATAATT-3′ (for R65A), 5′-GATCCAGGCAAGTTTTGGAGACATGTTTATGCATCACTTCTTCTTATCG-3′ and 5′-CGATAAGAAGAAGTGATGCATAAACATGTCTCCAAAACTTGCCTGGATC-3′ (for K78A), and 5′-GATTGTAGACATCATACTATGGAAATTAAAGCATTGGTTGAGTTCCAA-3′ and 5′-TTGGAACTCAACCAATGCTTTAATTTCCATAGTATGATGTCTACAATC-3′ (for T107A).
MBP:epsin binding assay
100 ml Dictyostelium suspension culture was harvested, washed in ice-cold binding buffer (20 mM piperazine-N,N′-bis[2-ethanesulfonic acid] pH 6.8, 1.5 mM EDTA, 15 mM MgCl2, 1 mM DTT and fungal protease inhibitor cocktail) (Vithalani et al., 1998) and resuspended to a concentration of 5×107 cells/ml. Cells were sonicated 5 times for 15 seconds at 50% power and centrifuged at 14,000 g for 20 minutes. MBP-epsin253-677 or MBP alone was purified from 1 l bacterial culture according to the manufacturer's protocol (see above), with the exception that the protein was not eluted from the amylose resin. 400 μl of beads were incubated with 1 ml prepared Dictyostelium lysate for 2 hours at 4°C with shaking. Beads were washed several times with cold binding buffer, and the bound fraction was eluted with hot RSB (reducing sample buffer). Samples were analyzed using standard immunoblotting protocols.
Lipid binding assay
1 nmol PtdIns or PtdIns(4,5)P2 (Echelon Biosciences, Salt Lake City, UT) was pipetted onto nitrocellulose membrane and allowed to air dry. The membrane was then blocked in ice-cold binding buffer with 3% dried milk for 30 minutes. Dictyostelium lysate was prepared as above and incubated with the membrane for 2 hours at 4°C with gentle shaking. The membrane was washed with 0.1% Tween-20 in PDF and then probed using standard immunoblotting protocols.
We thank members of the O'Halloran lab, Arturo De Lozanne, Janice Fischer and John Heuser for reading early versions of the manuscript. We are also thankful to Tobias Meyer for gift of the PH expression vector, and to John Heuser for advice, help and use of his TIRF microscope. This work is supported by an NSF Graduate Research Fellowship to R.J.B. and NIH RO1 GM048625 to T.J.O.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/121/??/????/DC1
- Accepted July 21, 2008.
- © The Company of Biologists Limited 2008