The Amino-terminal Domain of Desmoplakin Binds to Plakoglobin and Clusters Desmosomal Cadherin–Plakoglobin Complexes

The desmosome is a highly organized plasma membrane domain that couples intermediate filaments to the plasma membrane at regions of cell–cell adhesion. Desmosomes contain two classes of cadherins, desmogleins, and desmocollins, that bind to the cytoplasmic protein plakoglobin. Desmoplakin is a desmosomal component that plays a critical role in linking intermediate filament networks to the desmosomal plaque, and the amino-terminal domain of desmoplakin targets desmoplakin to the desmosome. However, the desmosomal protein(s) that bind the amino-terminal domain of desmoplakin have not been identified. To determine if the desmosomal cadherins and plakoglobin interact with the amino-terminal domain of desmoplakin, these proteins were co-expressed in L-cell fibroblasts, cells that do not normally express desmosomal components. When expressed in L-cells, the desmosomal cadherins and plakoglobin exhibited a diffuse distribution. However, in the presence of an amino-terminal desmoplakin polypeptide (DP-NTP), the desmosomal cadherins and plakoglobin were observed in punctate clusters that also contained DP-NTP. In addition, plakoglobin and DP-NTP were recruited to cell–cell interfaces in L-cells co-expressing a chimeric cadherin with the E-cadherin extracellular domain and the desmoglein-1 cytoplasmic domain, and these cells formed structures that were ultrastructurally similar to the outer plaque of the desmosome. In transient expression experiments in COS cells, the recruitment of DP-NTP to cell borders by the chimera required co-expression of plakoglobin. Plakoglobin and DP-NTP co-immunoprecipitated when extracted from L-cells, and yeast two hybrid analysis indicated that DP-NTP binds directly to plakoglobin but not Dsg1. These results identify a role for desmoplakin in organizing the desmosomal cadherin–plakoglobin complex and provide new insights into the hierarchy of protein interactions that occur in the desmosomal plaque.Desmosomes are highly organized adhesive intercellular junctions that couple intermediate filaments to the cell surface at sites of cell–cell adhesion (Farquhar and Palade, 1963; Staehelin, 1974; Schwarz et al., 1990; Garrod, 1993; Collins and Garrod, 1994; Cowin and Burke, 1996; Kowalczyk and Green, 1996). Desmosomes are prominent in tissues that experience mechanical stress, such as heart and epidermis, and the disruption of desmosomes or the intermediate filament system in these organs has devastating effects on tissue integrity (Steinert and Bale, 1993; Coulombe and Fuchs, 1994; Fuchs, 1994; McLean and Lane, 1995; Stanley, 1995; Bierkamp et al., 1996; Ruiz et al., 1996). Desmosomes are highly insoluble structures that can withstand harsh denaturing conditions (Skerrow and Matoltsy, 1974; Gorbsky and Steinberg, 1981; Jones et al., 1988; Schwarz et al., 1990). This property of desmosomes facilitated early identification of desmosomal components but has impaired subsequent biochemical analysis of the protein complexes that form between desmosomal components. Ultrastructurally, desmosomes contain a core region that includes the plasma membranes of adjacent cells and a cytoplasmic plaque that anchors intermediate filaments to the plasma membrane. The plaque can be further divided into an outer dense plaque subjacent to the plasma membrane and an inner dense plaque through which intermediate filaments appear to loop.Molecular genetic analysis has revealed that the desmosomal glycoproteins, the desmogleins and desmocollins, are members of the cadherin family of cell–cell adhesion molecules (for review see Buxton et al., 1993, 1994; Cowin and Mechanic, 1994; Kowalczyk et al., 1996). The classical cadherins, such as E-cadherin, mediate calcium-dependent, homophilic cell–cell adhesion (Nagafuchi et al., 1987). The mechanism by which the desmosomal cadherins mediate cell–cell adhesion remains elusive (Amagai et al., 1994; Chidgey et al., 1996; Kowalczyk et al., 1996), although heterophilic interactions have recently been detected between desmogleins and desmocollins (Chitaev and Troyanovsky, 1997). Both classes of the desmosomal cadherins associate with the cytoplasmic plaque protein plakoglobin (Kowalczyk et al., 1994; Mathur et al., 1994; Roh and Stanley, 1995b ; Troyanovsky et al., 1994), which is part of a growing family of proteins that share a repeated motif first identified in the Drosophila protein Armadillo (Peifer and Wieschaus, 1990). This multigene family also includes the desmosomal proteins band 6/plakophilin 1, plakophilin 2a and 2b, and p0071, which are now considered to comprise a subclass of the armadillo family of proteins (Hatzfeld et al., 1994; Heid et al., 1994; Schmidt et al., 1994; Hatzfeld and Nachtsheim, 1996; Mertens et al., 1996).The most abundant desmosomal plaque protein is desmoplakin, which is predicted to be a homodimer containing two globular end domains joined by a central α-helical coiled-coil rod domain (O''Keefe et al., 1989; Green et al., 1990; Virata et al., 1992). Previous studies have demonstrated that the carboxyl-terminal domain of desmoplakin interacts with intermediate filaments (Stappenbeck and Green, 1992; Stappenbeck et al., 1993; Kouklis et al., 1994; Meng et al., 1997), and the amino-terminal domain of desmoplakin is required for desmoplakin localization to the desmosomal plaque (Stappenbeck et al., 1993). Direct evidence supporting a role for desmoplakin in intermediate filament attachment to desmosomes was provided recently when expression of an amino-terminal polypeptide of desmoplakin was found to displace endogenous desmoplakin from cell borders and disrupt intermediate filament attachment to the cell surface in A431 epithelial cell lines (Bornslaeger et al., 1996).The classical cadherins, such as E-cadherin, bind directly to both β-catenin and plakoglobin (Aberle et al., 1994; Jou et al., 1995; for review see Cowin and Burke, 1996). β-Catenin is also an armadillo family member (McCrea et al., 1991; Peifer et al., 1992), and both plakoglobin and β-catenin bind directly to α-catenin (Aberle et al., 1994, 1996; Jou et al., 1995; Sacco et al., 1995; Obama and Ozawa, 1997). α-Catenin is a vinculin homologue (Nagafuchi et al., 1991) and associates with both α-actinin and actin (Knudson et al., 1995; Rimm et al., 1995; Nieset et al., 1997). Through interactions with β- and α-catenin, E-cadherin is coupled indirectly to the actin cytoskeleton, and this linkage is required for the adhesive activity of E-cadherin (Ozawa et al., 1990; Shimoyama et al., 1992). In addition, E-cadherin association with plakoglobin appears to be required for assembly of desmosomes (Lewis et al., 1997), underscoring the importance of E-cadherin in the overall program of intercellular junction assembly. However, the hierarchy of molecular interactions that couple the desmosomal cadherins to the intermediate filament cytoskeleton is largely unknown, although the desmocollin cytoplasmic domain appears to play an important role in recruiting components of the desmosomal plaque (Troyanovsky et al., 1993, 1994). Since desmosomal cadherins form complexes with plakoglobin and because the amino-terminal domain of desmoplakin is required for desmoplakin localization at desmosomes, we hypothesized that the amino-terminal domain of desmoplakin interacts with the desmosomal cadherin– plakoglobin complex.In previous studies, we used L-cell fibroblasts to characterize plakoglobin interactions with the cytoplasmic domains of the desmosomal cadherins and found that the desmosomal cadherins regulate plakoglobin metabolic stability (Kowalczyk et al., 1994) but do not mediate homophilic adhesion (Kowalczyk et al., 1996). To test the ability of the desmoplakin amino-terminal domain to interact with the desmosomal cadherin–plakoglobin complex, we established a series of L-cell lines expressing the desmosomal cadherins in the presence or absence of a desmoplakin amino-terminal polypeptide (DP-NTP).¹ The results indicate that one important function of the desmoplakin amino-terminal domain is to cluster desmosomal cadherin–plakoglobin complexes. In addition, DP-NTP and plakoglobin were found to form complexes that could be co-immunoprecipitated from L-cell lysates. Using the yeast two hybrid system, DP-NTP was found to bind directly to plakoglobin but not Dsg1. These data suggest that plakoglobin couples the amino-terminal domain of desmoplakin to the desmosomal cadherins and that desmoplakin plays an important role in organizing the desmosomal cadherin–plakoglobin complex into discrete plasma membrane domains.     

A Novel Class of RanGTP Binding Proteins

The importin-α/β complex and the GTPase Ran mediate nuclear import of proteins with a classical nuclear localization signal. Although Ran has been implicated also in a variety of other processes, such as cell cycle progression, a direct function of Ran has so far only been demonstrated for importin-mediated nuclear import. We have now identified an entire class of ∼20 potential Ran targets that share a sequence motif related to the Ran-binding site of importin-β. We have confirmed specific RanGTP binding for some of them, namely for two novel factors, RanBP7 and RanBP8, for CAS, Pse1p, and Msn5p, and for the cell cycle regulator Cse1p from Saccharomyces cerevisiae. We have studied RanBP7 in more detail. Similar to importin-β, it prevents the activation of Ran''s GTPase by RanGAP1 and inhibits nucleotide exchange on RanGTP. RanBP7 binds directly to nuclear pore complexes where it competes for binding sites with importin-β, transportin, and apparently also with the mediators of mRNA and U snRNA export. Furthermore, we provide evidence for a Ran-dependent transport cycle of RanBP7 and demonstrate that RanBP7 can cross the nuclear envelope rapidly and in both directions. On the basis of these results, we propose that RanBP7 might represent a nuclear transport factor that carries an as yet unknown cargo, which could apply as well for this entire class of related RanGTP-binding proteins.The nuclear pore complexes (NPC)¹ are the sites where the exchange of macromolecules between nucleus and cytoplasm occurs (Feldherr et al., 1984). Transport through the NPCs is bidirectional and comprises a multitude of substrates. Small molecules can passively diffuse through the NPC. The transport of proteins and RNAs >40–60 kD is, however, generally an active process, i.e., it is energy dependent (Newmeyer et al., 1986) and mediated by saturable transport receptors (Goldfarb et al., 1986; Michaud and Goldfarb, 1991; Jarmolowski et al., 1994). To accomplish multiple rounds of transport, these transport receptors are thought to shuttle between nucleus and cytoplasm (Goldfarb et al., 1986). An import receptor, for example, has to bind its import substrate initially in the cytoplasm, release it after NPC passage into the nucleus, and return to the cytoplasm without the cargo. Conversely, an export factor has to bind the export substrate only in the nucleus and on the way out. This model predicts asymmetry in these transport cycles and implies that the binding of the transport receptor to its cargo is regulated by the different environments of nucleus and cytoplasm.The nuclear import of proteins with a classical nuclear localization signal (NLS) is to date the best characterized nucleocytoplasmic transport pathway (for reviews see Görlich and Mattaj, 1996; Koepp and Silver, 1996; Schlenstedt, 1996). The signal contains one or more clusters of basic amino acids (for review see Dingwall and Laskey, 1991) and is recognized by the importin-α/β complex (for references see Sweet and Gerace, 1995; Panté and Aebi, 1996). The α subunit provides the NLS binding site (Imamoto et al., 1995; Weis et al., 1995) and is therefore also called the NLS receptor (Adam and Gerace, 1991). The β subunit accounts for the interaction with the NPC (Görlich et al., 1995; Moroianu et al., 1995) and carries importin-α with the NLS substrate into the nucleus. The translocation into the nucleus is terminated by the disassembly of the importin complex, and both subunits are returned probably separately to the cytoplasm. Importin-α interacts with -β via its importin-β binding domain (IBB domain; Görlich et al., 1996a ; Weis et al., 1996a ). Binding to importin-β with an IBB domain is sufficient for nuclear entry, and the IBB domain can therefore be regarded as the nuclear targeting signal of importin-α. The export domain of importin-α has not yet been identified, but it is distinct from the IBB domain.The small GTPase Ran (Drivas et al., 1990; Bischoff and Ponstingl, 1991b ; Belhumeur et al., 1993) plays a key role in NLS-dependent protein import (Melchior et al., 1993; Moore and Blobel, 1993). GTP hydrolysis by Ran is absolutely essential for import (Melchior et al., 1993; Moore and Blobel, 1993; Schlenstedt et al., 1995a ; Palacios et al., 1996) and is possibly even its sole source of energy (Weis et al., 1996b ). Although the molecular mechanism of import is far from being understood, it appears that Ran fulfils at least two distinct functions in this process: first, Ran''s GTP cycle probably drives translocation into the nucleus (Melchior et al., 1993; Moore and Blobel, 1993; Weis et al., 1996b ), which appears to involve the binding of (cytoplasmic) RanGDP to the NPC, followed by nucleotide exchange and GTP hydrolysis, but it does not involve binding of RanGTP to importin-β (Görlich et al., 1996b ). Unfortunately, nothing is known of how Ran''s GTP cycle would translate into a directed movement through the NPC. Secondly, Ran regulates the interaction between importin-α and -β (Rexach and Blobel, 1995; Chi et al., 1996; Görlich et al., 1996b ). Binding of RanGTP to importin-β disassembles the importin-α/β complex at the nuclear side of the NPC, thereby terminating translocation (Görlich et al., 1996b ). The asymmetric distribution of Ran''s principal GDP/GTP exchange factor (RCC1; Bischoff and Ponstingl, 1991a ) and GTPase activating protein (RanGAP1, or RNA1 in yeast; Bischoff et al., 1995a ; Becker et al., 1995) crucially determines where the importin heterodimer can form and where it is forced to dissociate. RCC1 is a nuclear, chromatin-bound protein (Ohtsubo et al., 1987, 1989) that generates RanGTP in the nucleus, whereas free RanGTP is depleted from the cytoplasm by RanGAP1, which is excluded from the nucleoplasm (Hopper et al., 1990; Matunis et al., 1996; Mahajan et. al, 1997). Thus, low RanGTP levels in the cytoplasm allow importin-α to bind -β, and the high RanGTP concentration in the nuclear compartment dissociates the importin complex. The concentration of free RanGTP can, in this model, be regarded as a marker for cytoplasmic identity (low RanGTP) and nuclear identity (high RanGTP), which is “sensed” by the Ran-binding site in importin-β.It is likely that at least some properties of importin-β are shared by the mediators of the other nucleocytoplasmic transport pathways. This is emphasized by the recent identification of the importin-β–related transportin (Pollard et al., 1996) as an import receptor recognizing the M9 domain, the nuclear targeting signal in hnRNP A1 (Michael et al., 1995), and of yeast transportin (Kap 104p) as an import receptor for mRNA binding proteins (Aitchison et al., 1996). Furthermore, importin-β or its NPC-binding domain cross-compete with other pathways, such as M9-dependent import, NES-mediated nuclear export, and the export of mRNA and U snRNA (Kutay et al., 1997). This would suggest that these other transport receptors share at least some binding sites at the NPC and take a similar path through the nuclear pore complex as importin-β.In addition to importin-β, a number of other Ran-binding proteins are detectable in eukaryotic cells, e.g., in overlay blots using Ran γ-[³²P]GTP as a probe. These can be grouped into two classes (Lounsbury et al., 1994, 1996): first, those with a RanBP1 homology domain including the Ran binding protein 1 (RanBP1) itself (Coutavas et al., 1993; Bischoff et al., 1995b ) and the nuclear pore protein RanBP2, which has four RanBP1 homology domains (Wu et al., 1995; Yokoyama et al., 1995). Their binding to Ran can be competed by RanBP1. Second, importin-β and so far unidentified protein(s) of ∼120 kD whose Ran-binding is competed by importin-β but not by excess of RanBP1 (Lounsbury et al., 1994, 1996). Both RanBP1 and importin-β inhibit the nucleotide exchange on RanGTP (Coutavas et al., 1993; Lounsbury et al., 1994, 1996; Bischoff et al., 1995b ; Görlich et al., 1996b ). However, they do not cross-compete with each other for Ran binding but instead bind to different, nonoverlapping sites on Ran (Chi et al., 1996; Kutay et al., 1997; Lounsbury and Macara, 1997). Another striking difference is that RanBP1 facilitates the activation of Ran''s GTPase by RanGAP1 (Beddow et al., 1995; Bischoff et al., 1995b ), whereas the importin-β/RanGTP complex is entirely GAP resistant (Floer and Blobel, 1996; Görlich et al., 1996b ).Although a direct involvement of Ran has so far only been demonstrated in the importin-dependent transport pathway, perturbations in the Ran system have severe effects on a great variety of cellular functions, such as RNA processing, RNA export, regulation of chromosome structure, cell cycle progression, initiation of replication, microtubule structure, etc. (for review see Dasso, 1993; Sazer, 1996). One could argue that these effects are all secondary consequences from an impaired NLS-dependent protein import. However, it is also possible that these defects are more direct and that eukaryotic cells contain many immediate targets of Ran function.Here we describe a novel superfamily of Ran-binding proteins, which includes about a dozen factors in yeast and probably even more in higher eukaryotes. The members of this superfamily share with importin-β an NH₂-terminal sequence motif that appears to account for RanGTP binding. Indeed we could confirm the interaction with Ran for the following factors: RanBP7 and RanBP8, two novel, related proteins described here, Cse1p, a cell cycle regulator in yeast, CAS, which is required for apoptosis in cultured human cells, and for Msn5p and Pse1p from yeast. Of these we have characterized RanBP7 and RanBP8 in more detail. Both resemble closely importin-β in their interaction with Ran, and both bind directly to nuclear pore complexes. RanBP7 can cross the nuclear membrane rapidly and in both directions. We provide evidence for a transport cycle in which RanBP7 first enters the nucleus, binds RanGTP inside the nucleus as a prerequisite for rapid re-export to the cytoplasm, after which the RanBP7/RanGTP complex becomes finally disassembled by the concerted action of RanBP1 and RanGAP1 in the cytoplasm. We propose that during these transport cycles, RanBP7 would normally carry an as yet unidentified cargo. This means, RanBP7 and possibly also the other members of the RanBP7/Cse1p/ importin-β superfamily could function as transport receptors that shuttle between nucleus and cytoplasm. RanBP7 and importin-β form an abundant, heterodimeric complex in the cytoplasm that appears to have a function different from nuclear import of proteins with a classical NLS. It might be a way to regulate either RanBP7 or importin-β function. Alternatively, the RanBP7/importin-β complex might constitute an import receptor with a substrate specificity different from that of the importin-α/β complex.     

Ii Chain Controls the Transport of Major Histocompatibility Complex Class II Molecules to and from Lysosomes

Major histocompatibility complex class II molecules are synthesized as a nonameric complex consisting of three αβ dimers associated with a trimer of invariant (Ii) chains. After exiting the TGN, a targeting signal in the Ii chain cytoplasmic domain directs the complex to endosomes where Ii chain is proteolytically processed and removed, allowing class II molecules to bind antigenic peptides before reaching the cell surface. Ii chain dissociation and peptide binding are thought to occur in one or more postendosomal sites related either to endosomes (designated CIIV) or to lysosomes (designated MIIC). We now find that in addition to initially targeting αβ dimers to endosomes, Ii chain regulates the subsequent transport of class II molecules. Under normal conditions, murine A20 B cells transport all of their newly synthesized class II I-A^b αβ dimers to the plasma membrane with little if any reaching lysosomal compartments. Inhibition of Ii processing by the cysteine/serine protease inhibitor leupeptin, however, blocked transport to the cell surface and caused a dramatic but selective accumulation of I-A^b class II molecules in lysosomes. In leupeptin, I-A^b dimers formed stable complexes with a 10-kD NH₂-terminal Ii chain fragment (Ii-p10), normally a transient intermediate in Ii chain processing. Upon removal of leupeptin, Ii-p10 was degraded and released, I-A^b dimers bound antigenic peptides, and the peptide-loaded dimers were transported slowly from lysosomes to the plasma membrane. Our results suggest that alterations in the rate or efficiency of Ii chain processing can alter the postendosomal sorting of class II molecules, resulting in the increased accumulation of αβ dimers in lysosome-like MIIC. Thus, simple differences in Ii chain processing may account for the highly variable amounts of class II found in lysosomal compartments of different cell types or at different developmental stages.The initiation of most immune responses requires antigen recognition by helper T lymphocytes. The antigen receptors on T cells can only recognize antigens as small peptides bound to major histocompatibility complex (MHC)¹ class II molecules at the surface of antigen presenting cells (Cresswell, 1994; Germain, 1994). The complexes between class II molecules and antigenic peptides are formed intracellularly somewhere along the endocytic pathway (Germain, 1994; Wolf and Ploegh, 1995). This process requires the internalization of protein antigen and its delivery to a site suitable for the generation of antigenic peptides. In addition, the peptides must be generated within, or transferred to, a site to which newly synthesized MHC class II molecules are delivered and rendered competent for peptide binding (Davidson et al., 1991).Invariant (Ii) chain plays a central role in controlling the intracellular transport of MHC class II (Cresswell, 1996). In the ER, Ii chain is synthesized as a trimer that complexes with three αβ dimers of MHC class II (Roche et al., 1991). Its NH₂-terminal cytoplasmic domain contains a wellknown targeting signal that directs class II–Ii chain complexes to endosomes after exit from the TGN (Bakke and Dobberstein, 1990; Lotteau et al., 1990; Neefjes et al., 1990; Odorizzi et al., 1994; Pieters et al., 1993). Once in endosomes, Ii chain is subjected to proteolysis by acid hydrolases (Roche and Cresswell, 1991). Degradation occurs in a stepwise fashion, resulting in the appearance of class II– bound NH₂-terminal intermediates containing the Ii chain cytoplasmic domain, membrane anchor, and parts of its luminal domain (Newcomb and Cresswell, 1993). The intermediates accumulate in the presence of protease inhibitors that interfere with Ii chain processing such as the serinecysteine protease inhibitor leupeptin, treatment with which can also block the transport of at least some class II haplotypes to the cell surface (Amigorena et al., 1995; Blum and Cresswell, 1988; Neefjes and Ploegh, 1992). How leupeptin inhibits surface appearance is unknown.In human cells, Ii chain degradation intermediates include a 21–22-kD fragment (designated LIP [leupeptininducible peptide]) and a 10–12-kD fragment (designated SLIP [small leupeptin-inducible peptide]) (Blum and Cresswell, 1988; Maric et al., 1994). In murine cells, only a 10– 12-kD fragment has been identified (Ii-p10) (Amigorena et al., 1995). Ii-p10 remains as a trimer associated with three αβ dimers and blocks the binding of antigenic peptides (Amigorena et al., 1995; Morton et al., 1995). It is thus likely that Ii-p10 includes a luminal region of Ii chain (designated CLIP) known to occupy the peptide binding groove of αβ dimers. Cleavage of Ii-p10 by a leupeptinsensitive protease causes its dissociation from αβ dimers, while leaving CLIP in the peptide binding groove. The removal of CLIP is favored at acidic pH but is additionally catalyzed by a second MHC gene product, HLA-DM (Sloan et al., 1995; Denzin and Cresswell, 1995; Karlsson et al., 1994; Roche, 1995). In mutant cells lacking HLA-DM, there is defective loading of antigenic peptides and the appearance of CLIP-αβ dimers on the plasma membrane (Mellins et al., 1994; Riberdy et al., 1992).The precise site(s) where these events occur remains unclear. In A20 B cells, a specialized population of endosome-like vesicles designated CIIV (for class II vesicles) represents a site through which a majority of newly synthesized class II molecules pass en route to the cell surface and a place where antigenic peptides bind αβ dimers of the I-A^d haplotype (Amigorena et al., 1994, 1995; Barnes and Mitchell, 1995). CIIV are physically distinct from the bulk of endosomes and lysosomes and contain at least some HLA-DM (Pierre et al., 1996). Despite the fact that most of the αβ dimers reaching CIIV are newly synthesized, CIIV contain little or no intact Ii chain (Amigorena et al., 1995). Thus, Ii chain–αβ complexes first may be delivered to endosomes where Ii chain is cleaved before being delivered to CIIV. That peptide loading can occur in CIIV has been demonstrated by experiments showing that leupeptin causes CIIV to transiently accumulate Ii-p10– containing complexes, which can then bind peptide (Amigorena et al., 1995).In human Epstein-Barr virus–transformed B lymphoblasts, most class II molecules have been localized to structures collectively designated MIIC (for MHC class II compartment) (Peters et al., 1991; Tulp et al., 1994; West et al., 1994). MIICs differ from CIIVs in that the latter contain endosomal but not lysosomal markers, while MIICs have most or all of the features of lysosomes (Peters et al., 1991, 1995; Pierre et al., 1996). Interestingly, the distribution of class II between endosomal (CIIV) and lysosomal (MIIC) compartments varies widely among cell types. Since lysosomes are classically defined as terminal degradative organelles (Kornfeld and Mellman, 1989), such variations may reflect differences in the rates at which class II is turned over in different cell types. On the other hand, MIICs also contain the bulk of HLA-DM and can host the loading of antigenic peptides onto class II molecules (Sanderson et al., 1994). The extent to which these complexes escape degradation and reach the cell surface is unclear. Nor is it at all clear how different cell types regulate the intracellular distribution of class II molecules between early and late endocytic compartments.We now show that murine A20 cells expressing endogenous I-A^d and transfected I-A^b normally localize little class II in lysosomes. Selective lysosomal accumulation of I-A^b αβ dimers can be induced after leupeptin treatment. Interestingly, I-A^b dimers, but not I-A^d dimers, are induced by leupeptin to form stable complexes with Ii-p10. Upon removal of the inhibitor, the Ii-p10 was removed and class II molecules were slowly transported from lysosomes to the cell surface. Thus, the rate of dissociation of Ii chain intermediates can regulate whether newly synthesized class II molecules are transported to the plasma membrane or to lysosomes.     

Tyrosine Phosphorylation at a Site Highly Conserved in the L1 Family of Cell Adhesion Molecules Abolishes Ankyrin Binding and Increases Lateral Mobility of Neurofascin

This paper presents evidence that a member of the L1 family of ankyrin-binding cell adhesion molecules is a substrate for protein tyrosine kinase(s) and phosphatase(s), identifies the highly conserved FIGQY tyrosine in the cytoplasmic domain as the principal site of phosphorylation, and demonstrates that phosphorylation of the FIGQY tyrosine abolishes ankyrin-binding activity. Neurofascin expressed in neuroblastoma cells is subject to tyrosine phosphorylation after activation of tyrosine kinases by NGF or bFGF or inactivation of tyrosine phosphatases with vanadate or dephostatin. Furthermore, both neurofascin and the related molecule Nr-CAM are tyrosine phosphorylated in a developmentally regulated pattern in rat brain. The FIGQY sequence is present in the cytoplasmic domains of all members of the L1 family of neural cell adhesion molecules. Phosphorylation of the FIGQY tyrosine abolishes ankyrin binding, as determined by coimmunoprecipitation of endogenous ankyrin and in vitro ankyrin-binding assays. Measurements of fluorescence recovery after photobleaching demonstrate that phosphorylation of the FIGQY tyrosine also increases lateral mobility of neurofascin expressed in neuroblastoma cells to the same extent as removal of the cytoplasmic domain. Ankyrin binding, therefore, appears to regulate the dynamic behavior of neurofascin and is the target for regulation by tyrosine phosphorylation in response to external signals. These findings suggest that tyrosine phosphorylation at the FIGQY site represents a highly conserved mechanism, used by the entire class of L1-related cell adhesion molecules, for regulation of ankyrin-dependent connections to the spectrin skeleton.Vertebrate L1, neurofascin, neuroglial cell adhesion molecule (Ng-CAM),¹ Ng-CAM–related cell adhesion molecule (Nr-CAM), and Drosophila neuroglian are members of a family of nervous system cell adhesion molecules that possess variable extracellular domains comprised of Ig and fibronectin type III domains and a relatively conserved cytoplasmic domain (Grumet, 1991; Hortsch and Goodman, 1991; Rathgen and Jessel, 1991; Sonderegger and Rathgen, 1992; Hortsch, 1996). Members of this family, including a number of alternatively spliced forms, are abundant in the nervous system during early development as well as in adults. Neurofascin and Nr-CAM, for example, constitute ∼0.5% of the total membrane protein in adult brain (Davis et al., 1993; Davis and Bennett, 1994). Cellular functions attributed to the L1 family include axon fasciculation (Stallcup and Beasley, 1985; Landmesser et al., 1988; Brummendorf and Rathjen, 1993; Bastmeyer et al., 1995; Itoh et al., 1995; Magyar-Lehmann et al., 1995), axonal guidance (van den Pol and Kim, 1993; Liljelund et al., 1994; Brittis and Silver, 1995; Brittis et al., 1995; Lochter et al., 1995; Wong et al., 1996), neurite extension (Chang et al., 1987; Felsenfeld et al., 1994; Hankin and Lagenaur, 1994; Ignelzi et al., 1994; Williams et al., 1994a ,b,c,d; Doherty et al., 1995; Zhao and Siu, 1995), a role in long term potentiation (Luthl et al., 1994), synaptogenesis (Itoh et al., 1995), and myelination (Wood et al., 1990). The potential clinical importance of this group of proteins has been emphasized by the findings that mutations in the L1 gene on the X chromosome are responsible for developmental anomalies including hydrocephalus and mental retardation (Rosenthal et al., 1992; Jouet et al., 1994; Wong et al., 1995).The conserved cytoplasmic domains of L1 family members include a binding site for the membrane skeletal protein ankyrin. This interaction was first described for neurofascin (Davis et. al., 1993) and subsequently has been observed for L1, Nr-CAM (Davis and Bennett, 1994), and Drosophila neuroglian (Dubreuil et al., 1996). The membrane-binding domain of ankyrin contains two distinct sites for neurofascin and has the potential to promote lateral association of neurofascin and presumably other L1 family members (Michaely and Bennett, 1995). Nodes of Ranvier are physiologically relevant axonal sites where ankyrin and L1 family members collaborate, based on findings of colocalization of a specialized isoform of ankyrin with alternatively spliced forms of neurofascin and NrCAM in adults (Davis et al., 1996) as well as in early axonal developmental intermediates (Lambert, S., J. Davis, P. Michael, and V. Bennett. 1995. Mol. Biol. Cell. 6:98a).L1, after homophilic and/or heterophilic binding, participates in signal transduction pathways that ultimately are associated with neurite extension and outgrowth (Ignelzi et al., 1994; Williams et al., 1994a ,b,c,d; Doherty et al., 1995). L1 copurifies with a serine–threonine protein kinase (Sadoul et al., 1989) and is phosphorylated on a serine residue that is not conserved among other family members (Wong et al., 1996). L1 pathway(s) may also involve G proteins, calcium channels, and tyrosine phosphorylation (Williams et al., 1994a ,b,c,d; Doherty et al., 1995). After homophilic interactions, L1 directly activates a tyrosine signaling cascade after a lateral association of its ectodomain with the fibroblast growth factor receptor (Doherty et al., 1995). Antibodies against L1 have also been shown to activate protein tyrosine phosphatase activity in growth cones (Klinz et al., 1995). However, details of the downstream substrates of L1-promoted phosphorylation and dephosphorylation and possible roles of the cytoplasmic domain are not known.Tyrosine phosphorylation is well established to modulate cell–cell and cell–extracellular matrix interactions involving integrins and their associated proteins (Akiyama et al., 1994; Arroyo et al., 1994; Schlaepfer et al., 1994; Law et al., 1996) as well as the cadherins (Balsamo et al., 1996; Krypta et al., 1996; Brady-Kalnay et al., 1995; Shibamoto et al., 1995; Hoschuetzky et al., 1994; Matsuyoshi et al., 1992). For example, the adhesive functions of the calciumdependent cadherin cell adhesion molecule are mediated by a dynamic balance between tyrosine phosphorylation of β-catenin by TrkA and dephosphorylation via the LARtype protein tyrosine phosphatase (Krypta et al., 1996). In this example the regulation of binding among the structural proteins is the result of a coordination between classes of protein kinases and protein phosphatases.This study presents evidence that neurofascin, expressed in a rat neuroblastoma cell line, is a substrate for both tyrosine kinases and protein tyrosine phosphatases at a tyrosine residue conserved among all members of the L1 family. Site-specific tyrosine phosphorylation promoted by both tyrosine kinase activators (NGF and bFGF) and protein tyrosine phosphatase inhibitors (dephostatin and vanadate) is a strong negative regulator of the neurofascin– ankyrin binding interaction and modulates the membrane dynamic behavior of neurofascin. Furthermore, neurofascin and, to a lesser extent Nr-CAM, are also shown here to be tyrosine phosphorylated in developing rat brain, implying a physiological relevance to this phenomenon. These results indicate that neurofascin may be a target for the coordinate control over phosphorylation that is elicited by protein kinases and phosphatases during in vivo tyrosine phosphorylation cascades. The consequent decrease in ankyrin-binding capacity due to phosphorylation of neurofascin could represent a general mechanism among the L1 family members for regulation of membrane–cytoskeletal interactions in both developing and adult nervous systems.     

Novel Cytokine-independent Induction of Endothelial Adhesion Molecules Regulated by Platelet/Endothelial Cell Adhesion Molecule (CD31)

Tumor necrosis factor–α, interleukin-1, and endotoxin stimulate the expression of vascular endothelial cell (EC) adhesion molecules. Here we describe a novel pathway of adhesion molecule induction that is independent of exogenous factors, but which is dependent on integrin signaling and cell–cell interactions. Cells plated onto gelatin, fibronectin, collagen or fibrinogen, or anti-integrin antibodies, expressed increased amounts of E-selectin, vascular cell adhesion molecule–1, and intercellular adhesion molecule–1. In contrast, ECs failed to express E-selectin when plated on poly-l-lysine or when plated on fibrinogen in the presence of attachment-inhibiting, cyclic Arg-Gly-Asp peptides. The duration and magnitude of adhesion molecule expression was dependent on EC density. Induction of E-selectin on ECs plated at confluent density was transient and returned to basal levels by 15 h after plating when only 7 ± 2% (n = 5) of cells were positive. In contrast, cells plated at low density displayed a 17-fold greater expression of E-selectin than did high density ECs with 57 ± 4% (n = 5) positive for E-selectin expression 15 h after plating, and significant expression still evident 72 h after plating. The confluency-dependent inhibition of expression of E-selectin was at least partly mediated through the cell junctional protein, platelet/endothelial cell adhesion molecule–1 (PECAM-1). Antibodies against PECAM-1, but not against VE-cadherin, increased E-selectin expression on confluent ECs. Co– culture of subconfluent ECs with PECAM-1– coated beads or with L cells transfected with full-length PECAM-1 or with a cytoplasmic truncation PECAM-1 mutant, inhibited E-selectin expression. In contrast, untransfected L cells or L cells transfected with an adhesion-defective domain 2 deletion PECAM-1 mutant failed to regulate E-selectin expression. In an in vitro model of wounding the wound front displayed an increase in the number of E-selectin–expressing cells, and also an increase in the intensity of expression of E-selectin positive cells compared to the nonwounded monolayer. Thus we propose that the EC junction, and in particular, the junctional molecule PECAM-1, is a powerful regulator of endothelial adhesiveness.The endothelial lining of the vascular system normally displays a nonactivated, nonadhesive phenotype. Stimulation with agents such as tumor necrosis factor-α (TNF-α)¹, interleukin-1 (IL-1), or lipopolysaccharide (LPS) are known to induce the expression of proteins on the endothelial surface that mediate coagulation (Bevilacqua et al., 1986), leukocyte adhesion (Bevilacqua et al., 1985; Gamble et al., 1985; Pober et al., 1986b ; Doherty et al., 1989), and leukocyte transendothelial migration (Furie et al., 1989; Moser et al., 1989). The endothelial antigens that are important for the adhesion of leukocytes are members of the selectin family, E- and P-selectin, and the immunoglobulin gene superfamily, vascular cell adhesion molecule–1 (VCAM-1) and intercellular adhesion molecule–1 (ICAM-1) (Carlos and Harlan, 1994; Litwin et al., 1995).The induction of E-selectin expression on endothelial cells (ECs) in vitro after cytokine stimulation is transient and independent of the continued presence of the stimulant (Pober et al., 1986a ). Previous studies have shown that E-selectin mRNA and protein levels peak between 2 and 4 h, respectively, after treatment with an agonist, returning to near basal levels by 24 h (Bevilacqua et al., 1989; Read et al., 1994). VCAM-1 (Osborn et al., 1989) and ICAM-1 (Pober et al., 1986b ) are maximal 6 and 12 h, respectively, after stimulation.In contrast to the transiency of E-selectin and VCAM expression demonstrated by the in vitro data, these antigens have been detected on venular endothelium in chronic inflammatory lesions, such as the synovium in rheumatoid arthritis (Koch et al., 1991), and the skin in psoriasis (Petzelbauer et al., 1994). E-selectin expression is also detected on angiogenic vessels in human hemangiomas, a noninflammatory angiogenic disease (Kraling et al., 1996). Moreover, the architecture and anatomic localization of capillary loops influence the pattern of endothelial expression of E-selectin and VCAM-1, independently of the availability of cytokines (Petzelbauer et al., 1994). Thus it is likely that alternate control mechanisms exist to allow prolonged, locality-based expression of adhesion molecules on the endothelium. At least one of these alternate mechanisms may be flow, since increased shear stress has been shown to selectively modulate adhesion molecule expression, upregulating ICAM-1 but not E-selectin or VCAM-1 (Nagel et al., 1994).Since sites of inflammation are often associated with morphological changes including cell retraction of the endothelium (Schumacher, 1973), we hypothesized that cell contacts may be important in the regulation of endothelial phenotype. We describe here the central role of the junctional protein, platelet/endothelial cell adhesion molecule–1 (PECAM-1), through the formation of cell–cell interactions, in the maintenance of the functional integrity of the endothelial monolayer. Furthermore, we demonstrate a novel pathway for the induction of adhesion molecules on endothelial cells that is independent of exogenous addition of cytokines, but is related to integrin- and cell shape–associated signaling events.     

Neurotrophin-3–enhanced Nerve Regeneration Selectively Improves Recovery of Muscle Fibers Expressing Myosin Heavy Chains 2b

The purpose of this study was to evaluate the effect of neurotrophin 3 (NT-3) enhanced nerve regeneration on the reinnervation of a target muscle. Muscle fibers can be classified according to their mechanical properties and myosin heavy chain (MHC) isoform composition. MHC1 containing slow-type and MHC2a or 2b fast-type fibers are normally distributed in a mosaic pattern, their phenotype dictated by motor innervation. After denervation, all fibers switch to fast-type MHC2b expression and also undergo atrophy resulting in loss of muscle mass. After regeneration, discrimination between fast and slow fibers returns, but the distribution and fiber size change according to the level of reinnervation. In this study, rat gastrocnemius muscles (ipsilateral and contralateral to the side of nerve injury) were collected up to 8 mo after nerve repair, with or without local delivery of NT-3. The phenotype changes of MHC1, 2a, and 2b were analyzed by immunohistochemistry, and fiber type proportion, diameter, and grouping were assessed by computerized image analysis. At 8 mo, the local delivery of NT-3 resulted in significant improvement in gastrocnemius muscle weight compared with controls (NT-3 group 47%, controls 39% weight of contralateral normal muscle; P < 0.05). NT-3 delivery resulted in a significant increase in the proportion (NT-3 43.3%, controls 35.7%; P < 0.05) and diameter (NT-3 87.8 μm, controls 70.8 μm; P < 0.05) of fast type 2b fibers after reinnervation. This effect was specific to type 2b fibers; no normalization was seen in other fiber types.This study indicates that NT-3–enhanced axonal regeneration has a beneficial effect on the motor target organ. Also, NT-3 may be specifically affecting a subset of motoneurons that determine type 2b muscle fiber phenotype. As NT-3 was topically applied to cut nerves, our data suggest a discriminating effect of the neurotrophin on neuro–muscular interaction. These results would imply that muscle fibers may be differentially responsive to other neurotrophic factors and indicate the potential clinical role of NT-3 in the prevention of muscle atrophy after nerve injury.There has been much recent interest in the use of growth factors to augment peripheral nerve regeneration. A family of growth factors collectively known as the neurotrophins are now considered critical for the development, maintenance, and regeneration of the nervous system. The neurotrophin family includes NGF, brain derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3),¹ and neurotrophin-4/5 (NT-4/5) (Lewin and Barde, 1996; Lindsay, 1996). Little is known of their effect on regeneration of the peripheral nervous system.NT-3 has been shown to act on a subpopulation of muscle sensory neurons innervating muscle spindles and Golgi tendon organs, and there is also evidence of its effect on a subpopulation of cutaneous afferents (Ernfors et al., 1994; Tessarolo et al., 1994; Airaksinen et al., 1996). NT-3 has shown various effects on motor nerve regeneration, including differentiation of motoneurons from avian neural tube progenitor cells (Averbuch-Heller et al., 1994) and survival of neonatal and adult motoneurons in vitro (Hughes et al., 1993) and of neonatal motoneurons in vivo (Li et al., 1994; Vejsada et al., 1995), although the evidence is sometimes contradictory (Eriksson et al., 1994). In cocultures of adult muscle and embryonic motoneurons, NT-3 enhances the number and length of neurite outgrowths, the density of endplates per muscle fiber, and the amount of muscle innervation (Braun et al., 1996). NT-3 also plays a role in functional maturation of neuromuscular synapses (Lohoff et al., 1993; Wang et al., 1995) and regulates the cholinergic phenotype of developing motoneurons (Wong et al., 1993; Kato and Lindsay, 1994). NT-3 knockout mice show a loss of all muscle spindle afferent innervation and fusimotor neurons to the muscle but lose only few skeletomotor nerve fibers (Kucera et al., 1995a ). About 80% of adult motoneurons express the NT-3–specific trkC receptor (Henderson et al., 1993; Griesbeck et al., 1995), and NT-3 is the predominant neurotrophin expressed in skeletal muscle (Griesbeck et al., 1995). Furthermore, NT-3 is internalized and retrogradely transported from the periphery to motoneuron cell bodies (Di Stefano et al., 1992). Thus, there is experimental and circumstantial evidence to suggest that NT-3 may play a role in adult motoneurons, although in vivo data on the survival effect of NT-3 on adult motoneurons is still lacking. Furthermore, there is no evidence of an NT-3–dependent effect on neuro–muscular interaction.When a skeletal muscle is denervated and subsequently reinnervated, a characteristic sequence of events ensues. The muscle rapidly loses weight as the muscle fibers atrophy (Pellegrino and Franzini, 1963), but after reinnervation, it gradually recovers mass to a variable extent, depending upon the degree of reinnervation (Bertelli and Mira, 1995) and correlating with the maximum force of contraction (Gillespie et al., 1987). The fibers within an individual skeletal muscle do not exist as a homogenous population but can be classified according to their different metabolic and contractile properties (Burke et al., 1971; Peter et al., 1972). Also, they can be identified morphologically according to differential expression of specific myosin heavy chain isoforms. Slow, oxidative type 1 muscle fibers contain myosin heavy chain 1 (MHC 1), fast oxidative glycolytic type 2a fibers contain myosin heavy chain 2a (MHC 2a), while fast glycolytic type 2b fibers contain myosin heavy chain 2b (MHC 2b) (Bar and Pette, 1988). Muscle fiber phenotype is conferred by its innervation, and changes of neuro–muscular interaction lead to alteration of muscle fiber phenotype (Romanul and Van der Meulen, 1966; Fex and Sonneson, 1970; Salmons and Sreter, 1975). The relative proportions of fiber types vary with age, sex, strain, species, and muscle type (Maltin et al., 1989). Generally, there is a high proportion of type 1 fibers in postural muscles (e.g., soleus) and of type 2 fibers in fast muscles (e.g., extensor digitorum longus), while in mixed muscles (e.g., gastrocnemius) there are varying proportions of each type. Muscle fiber type proportion also varies dynamically with physiological and pathological parameters (Jansson et al., 1978; Green et al., 1983; Izumo et al., 1986; Goldspink et al., 1992; Pette and Vrbova, 1992). For example, the distribution of fiber types in normal muscle is dispersed in a “mosaic pattern,” but after denervation and reinnervation of the muscle there is a shift to grouping (Karpati and Engel, 1968). Also, there is a change in the proportions of fiber types, and in the rat the majority of fibers become initially fast with denervation, with subsequent fiber specialization being dictated by patterns of reinnervation (Fields and Ellisman, 1986). This plastic nature of muscle makes it an interesting model to investigate reinnervation changes that may occur after NT-3 administration.We have recently demonstrated that the local delivery of NT-3 to rat sciatic nerve enhances the rate and amount of nerve regeneration, and at 8 mo postoperative, there was a 40% increase in the myelinated fiber count (Sterne et al., 1997). However, enhanced regeneration by itself is not necessarily indicative of a beneficial effect on the target muscles, such as reacquisition of more normal physiological function. Therefore, the aim of this study was to assess whether NT-3–enhanced nerve regeneration resulted in biochemical or morphological changes in a target muscle (gastrocnemius), which would be suggestive of significant improvement of physiological function above that seen after nerve repair without administration of neurotrophin. Immunohistochemistry, in conjunction with computerized quantification and morphometrical analysis, was used to analyze the number, size, and pattern of distribution of the MHC fiber types after denervation and reinnervation of the gastrocnemius muscle.     

ERS-24, a Mammalian v-SNARE Implicated in Vesicle Traffic between the ER and the Golgi

We report the identification and characterization of ERS-24 (Endoplasmic Reticulum SNARE of 24 kD), a new mammalian v-SNARE implicated in vesicular transport between the ER and the Golgi. ERS24 is incorporated into 20S docking and fusion particles and disassembles from this complex in an ATP-dependent manner. ERS-24 has significant sequence homology to Sec22p, a v-SNARE in Saccharomyces cerevisiae required for transport between the ER and the Golgi. ERS-24 is localized to the ER and to the Golgi, and it is enriched in transport vesicles associated with these organelles.Newly formed transport vesicles have to be selectively targeted to their correct destinations, implying the existence of a set of compartment-specific proteins acting as unique receptor–ligand pairs. Such proteins have now been identified (Söllner et al., 1993a ; Rothman, 1994): one partner efficiently packaged into vesicles, termed a v-SNARE,¹ and the other mainly localized to the target compartment, a t-SNARE. Cognate pairs of v- and t-SNAREs, capable of binding each other specifically, have been identified for the ER–Golgi transport step (Lian and Ferro-Novick, 1993; Søgaard et al., 1994), the Golgi–plasma membrane transport step (Aalto et al., 1993; Protopopov et al., 1993; Brennwald et al., 1994) in Saccharomyces cerevisiae, and regulated exocytosis in neuronal synapses (Söllner et al., 1993a ; for reviews see Scheller, 1995; Südhof, 1995). Additional components, like p115, rab proteins, and sec1 proteins, appear to regulate vesicle docking by controlling the assembly of SNARE complexes (Søgaard et al., 1994; Lian et al., 1994; Sapperstein et al., 1996; Hata et al., 1993; Pevsner et al., 1994).In contrast with vesicle docking, which requires compartment-specific components, the fusion of the two lipid bilayers uses a more general machinery derived, at least in part, from the cytosol (Rothman, 1994), which includes an ATPase, the N-ethylmaleimide–sensitive fusion protein (NSF) (Block et al., 1988; Malhotra et al., 1988), and soluble NSF attachment proteins (SNAPs) (Clary et al., 1990; Clary and Rothman, 1990; Whiteheart et al., 1993). Only the assembled v–t-SNARE complex provides high affinity sites for the consecutive binding of three SNAPs (Söllner et al., 1993b ; Hayashi et al., 1995) and NSF. When NSF is inactivated in vivo, v–t-SNARE complexes accumulate, confirming that NSF is needed for fusion after stable docking (Søgaard et al., 1994).The complex of SNAREs, SNAPs, and NSF can be isolated from detergent extracts of cellular membranes in the presence of ATPγS, or in the presence of ATP but in the absence of Mg²⁺, and sediments at ∼20 Svedberg (20S particle) (Wilson et al., 1992). In the presence of MgATP, the ATPase of NSF disassembles the v–t-SNARE complex and also releases SNAPs. It seems likely that this step somehow initiates fusion.To better understand vesicle flow patterns within cells, it is clearly of interest to identify new SNARE proteins. Presently, the most complete inventory is in yeast, but immunolocalization is difficult in yeast compared with animal cells, and many steps in protein transport have been reconstituted in animal extracts (Rothman, 1992) that have not yet been developed in yeast. Therefore, it is important to create an inventory of SNARE proteins in animal cells. The most unambiguous and direct method for isolating new SNAREs is to exploit their ability to assemble together with SNAPs and NSF into 20S particles and to disassemble into subunits when NSF hydrolyzes ATP. Similar approaches have already been successfully used to isolate new SNAREs implicated in ER to Golgi (Søgaard et al., 1994) and intra-Golgi transport (Nagahama et al., 1996), in addition to the original discovery of SNAREs in the context of neurotransmission (Söllner et al., 1993a ).Using this method, we now report the isolation and detailed characterization of ERS-24 (Endoplasmic Reticulum SNARE of 24 kD), a new mammalian v-SNARE that is localized to the ER and Golgi. ERS-24 is found in transport vesicles associated with the transitional areas of the ER and with the rims of Golgi cisternae, suggesting a role for ERS-24 in vesicular transport between these two compartments.     

The Yeast Adaptor Protein Complex,AP-3, Is Essential for the Efficient Delivery of Alkaline Phosphatase by the Alternate Pathway to the Vacuole

A novel clathrin adaptor-like complex, adaptor protein (AP)-3, has recently been described in yeast and in animals. To gain insight into the role of yeast AP-3, a genetic strategy was devised to isolate gene products that are required in the absence of the AP-3 μ chain encoded by APM3. One gene identified by this synthetic lethal screen was VPS45. The Vps pathway defines the route that several proteins, including carboxypeptidase Y, take from the late Golgi to the vacuole. However, vacuolar alkaline phosphatase (ALP) is transported via an alternate, intracellular route. This suggested that the apm3-Δ vps45 synthetic phenotype could be caused by a block in both the alternate and the Vps pathways. Here we demonstrate that loss of function of the AP-3 complex results in slowed processing and missorting of ALP. ALP is no longer localized to the vacuole membrane by immunofluorescence, but is found in small punctate structures throughout the cell. This pattern is distinct from the Golgi marker Kex2p, which is unaffected in AP-3 mutants. We also show that in the apm3-Δ mutant some ALP is delivered to the vacuole by diversion into the Vps pathway. Class E vps mutants accumulate an exaggerated prevacuolar compartment containing membrane proteins on their way to the vacuole or destined for recycling to the Golgi. Surprisingly, in AP-3 class E vps double mutants these proteins reappear on the vacuole. We suggest that some AP-3–dependent cargo proteins that regulate late steps in Golgi to vacuole transport are diverted into the Vps pathway allowing completion of transfer to the vacuole in the class E vps mutant.The formation of vesicles for transport between membrane-bound organelles requires assembly of coat proteins that are recruited from the cytosol. These proteins direct the sequestration and concentration of cargo as well as invagination of the membrane. One of the best studied classes of coats involved in vesicle budding is comprised of clathrin and its adaptor proteins (APs)¹, AP-1 and AP-2 (Schmid, 1997). In clathrin-mediated vesicle transport the AP complexes play the dual role of cargo selection and recruitment of clathrin to the membrane. These adaptors are heterotetramers containing two large chains (adaptins, α or γ and β), one medium chain (μ), and one small chain (σ). AP-1 (γ, β1, μ1, and σ1) functions in sorting at the TGN, whereas AP-2 (α, β2, μ2, and σ2) is involved in receptor capture at the PM during endocytosis.Although there is a great deal of evidence supporting the involvement of adaptors in clathrin-mediated vesicle budding, recent studies in animal cells have led to the discovery of a novel adaptor-like complex, AP-3, that seems to function independently of clathrin (Newman et al., 1995; Simpson et al., 1996). AP-3 has identical subunit architecture to AP-1 and AP-2, with two adaptin-like subunits (δ and β3), a medium chain (μ3), and a small chain (σ3) (Simpson et al., 1996, 1997; Dell''Angelica et al., 1997a , b ). AP-3 antibodies label a perinuclear region, perhaps the TGN, and punctate structures extending to the cell periphery, which may be endosomal compartments (Simpson et al., 1996, 1997; Dell''Angelica et al., 1997a ). However, the mammalian AP-3 complex does not colocalize with clathrin or AP-1 and AP-2 adaptors in cells and it does not copurify with brain clathrin-coated vesicles (Newman et al., 1995; Simpson et al., 1996, 1997; Dell''Angelica et al., 1997b ). Clues to the function of AP-3 have come from the discovery that the garnet gene of Drosophila encodes a protein closely related to δ adaptin (Ooi et al., 1997; Simpson et al., 1997). Mutations in garnet cause decreased pigmentation of the eyes and other tissues and a reduced number of pigment granules, which may be lysosome-like organelles (Ooi et al., 1997; Simpson et al., 1997). Thus, AP-3 is proposed to function in clathrin-independent transport between the TGN, endosomes and/or lysosomes, although its exact sorting function is still not known.Over the last several years, yeast homologues of the mammalian adaptor subunits have been identified, allowing for the examination of specific functions of these proteins in a genetically tractable organism. Genes encoding subunits sufficient for at least three complete AP complexes have been identified by sequence homology (Phan et al., 1994; Rad et al., 1995; Stepp et al., 1995) or by function (Panek et al., 1997). APL1-APL6 encode large chain/ adaptin-related subunits, APM1-APM4 encode μ-like chains, and APS1-APS3 are genes for σ-related proteins. Apl2p (β), Apl4p (γ), Apm1p (μ1), and Aps1p (σ1) are thought to be subunits of an AP-1–like complex that functions with clathrin at the late Golgi/TGN (Phan et al., 1994; Rad et al., 1995; Stepp et al., 1995; Payne, G., personal communication). Mutations in the yeast AP-1 genes enhance the growth and the α-factor processing defects of a temperature sensitive (ts) allele of the clathrin heavy chain gene (Phan et al., 1994; Rad et al., 1995; Stepp et al., 1995; Payne, G., personal communication). The latter phenotype is a hallmark of clathrin-deficient yeast, in which late Golgi/ TGN proteins, such as the α-factor processing enzymes Kex2p and dipeptidyl amino peptidase-A (DPAP)-A, are not retained in the late Golgi but escape to the cell surface (Seeger and Payne, 1992b ). To date, no yeast adaptor subunit has been shown to be important for endocytosis, although Apl3p, Apm4p, and Aps2p are most homologous to mammalian AP-2 α, μ2 and σ2, respectively.Recently, a yeast adaptor related to AP-3 of animal cells was described (Panek et al., 1997). It is comprised of Apl5p, Apl6p, Apm3p, and Aps3p, which show preferential homology to mammalian δ, β3, μ3, and σ3, respectively. Mutations in each of these subunits were isolated by their ability to suppress the lethality resulting from loss of function of PM casein kinase 1 encoded by a gene pair, YCK1 and YCK2. Yck activity was found to be required for constitutive endocytosis of the a-factor receptor (Ste3p), and AP-3 subunit mutations partially rescued this internalization defect (Panek et al., 1997). However, the AP complex itself is not necessary for endocytosis, nor is it required for sorting of carboxypeptidase Y (CPY) or retention of late Golgi proteins. Furthermore, unlike disruption of the yeast AP-1 complex, loss of AP-3 function causes no synthetic phenotype in combination with chc1 mutations, suggesting it may function independently of clathrin. Although these data indicated that Apl5p, Apl6p, Apm3p, and Aps3p comprise an AP-3-like adaptor, its precise sorting role was still not known.In this report we describe a genetic approach to determine the function of the yeast AP-3 complex. A colony sectoring screen was performed to identify genes that are essential in the absence of Apm3p, the yeast AP-3 μ chain. Such synthetic lethal screens can be used to identify functional homologues, genes whose proteins function in intersecting or parallel pathways, and genes whose proteins physically interact (Bender and Pringle, 1991). We have cloned the gene for the apm three synthetic lethal mutant, mts1-1, and found it encodes Vps45p, a protein involved in vacuolar protein sorting (Vps; Cowles et al., 1994; Piper et al., 1994). The Vps pathway is defined by >40 complementation groups whose proteins are required for the transport of a number of soluble and membrane-bound proteins, including CPY, protease A (PrA), and carboxypeptidase S (CPS) from the late Golgi/TGN to the vacuole (Stack et al., 1995; Cowles et al., 1997). This pathway is also essential for proper assembly of the vacuolar ATPase (Raymond et al., 1992). However, the type II vacuolar membrane protein alkaline phosphatase (ALP) follows an alternate intracellular pathway to the vacuole (Raymond et al., 1992; Nothwehr et al., 1995; Cowles et al., 1997; Piper et al., 1997). Few vps mutants prevent localization of ALP to the vacuolar membrane and its arrival at the vacuole is not dependent upon transport through the cell surface. The requirement for Apm3p in the absence of Vps45p suggested the possibility that at least one of these routes to the vacuole must be functional for survival and led us to examine ALP sorting in the AP-3 mutants. We show here that yeast AP-3 is essential for the transport of ALP via the alternative pathway to the vacuole.     

Three-dimensional Structure of Acanthamoeba castellanii Myosin-IB (MIB) Determined by Cryoelectron Microscopy of Decorated Actin Filaments

The Acanthamoeba castellanii myosin-Is were the first unconventional myosins to be discovered, and the myosin-I class has since been found to be one of the more diverse and abundant classes of the myosin superfamily. We used two-dimensional (2D) crystallization on phospholipid monolayers and negative stain electron microscopy to calculate a projection map of a “classical” myosin-I, Acanthamoeba myosin-IB (MIB), at ∼18 Å resolution. Interpretation of the projection map suggests that the MIB molecules sit upright on the membrane. We also used cryoelectron microscopy and helical image analysis to determine the three-dimensional structure of actin filaments decorated with unphosphorylated (inactive) MIB. The catalytic domain is similar to that of other myosins, whereas the large carboxy-terminal tail domain differs greatly from brush border myosin-I (BBM-I), another member of the myosin-I class. These differences may be relevant to the distinct cellular functions of these two types of myosin-I. The catalytic domain of MIB also attaches to F-actin at a significantly different angle, ∼10°, than BBM-I. Finally, there is evidence that the tails of adjacent MIB molecules interact in both the 2D crystal and in the decorated actin filaments.The myosin superfamily consists of at least 12 distinct classes that vary both in the sequence of their conserved myosin catalytic domains as well as in their unique tails (Mooseker and Cheney, 1995; Sellers and Goodson, 1995). For many years the only known myosins were the double-headed, filament-forming myosins found in muscle (conventional myosins or myosins-II). The remaining classes of myosin have been termed “unconventional myosins” to differentiate them from the myosins-II. Probably the most thoroughly studied class of unconventional myosins is the myosin-I class. These small, single-headed myosins bind to membrane lipids through a basic domain in their tail (for review see Pollard et al., 1991; Mooseker and Cheney, 1995). The first unconventional myosin (and first myosin-I) was isolated from Acanthamoeba castellanii (Pollard and Korn, 1973 a,b), and was purified on the basis of its K⁺, EDTA, and actin-activated MgATPase activities. However, this myosin was unusual in that it had a single heavy chain of ∼140 kD, in contrast to the two ∼200-kD heavy chains of myosin-II (Pollard and Korn, 1973 a).Three isoforms of the classical Acanthamoeba myosins-I are now known: myosins-IA, -IB, and -IC (Maruta and Korn, 1977a ,b; Maruta et al., 1979). Each of the isoforms consists of a conserved myosin catalytic domain, a binding site for one or two light chains, a basic domain, a GPA¹(Q) domain (rich in glycine, proline and alanine [glutamine]), and an scr-homology domain-3 (SH3) domain (Pollard et al., 1991; Mooseker and Cheney, 1995). These myosins-I can associate with membranes or with actin filaments through their tail domains. An electrostatic association of myosin-I with anionic phospholipids and with base-stripped membranes has been shown to occur (Adams and Pollard, 1989; Miyata et al., 1989; Hayden et al., 1990), and this interaction has been mapped to the basic domain (Doberstein and Pollard, 1992). Interestingly, these myosins also contain a second, ATP-insensitive actin binding site (Lynch et al., 1986) enabling them to mediate actin–actin movements (Albanesi et al., 1985; Fujisaki et al., 1985). In myosin-IA (Lynch et al., 1986) and myosin-IC (Doberstein and Pollard, 1992), this binding site was localized to the GPA domain. Acanthamoeba myosins-I have maximal steady-state actin-activated ATPase rates of ∼10–20 s⁻¹ (Pollard and Korn, 1973 b; Albanesi et al., 1983), and an unusual triphasic dependence upon actin concentration (Pollard and Korn, 1973 b; Albanesi et al., 1983). This triphasic activation is due to the actin cross-linking ability imparted by the ATP-insensitive actin binding site on the tail (Albanesi et al., 1985). Analysis of the individual steps in the ATPase cycle by transient kinetics revealed that the mechanism of myosin-IA is similar to slow skeletal muscle myosin, whereas myosin-IB (MIB) is similar to fast skeletal muscle myosin (Ostap and Pollard, 1996). The in vitro motility of myosin-I has also been well characterized (Zot et al., 1992). The maximal rate of filament sliding is ∼0.2 μm s⁻¹. Interestingly, this rate is ∼10–50× slower than the rates observed for skeletal muscle myosin, even though the ATPase rates are comparable.MIB consists of a 125-kD heavy chain and a single 27-kD light chain (Maruta et al., 1979; Jung et al., 1987). This isoform is primarily associated with the plasma membrane as well as vacuolar membranes (Baines et al., 1992). MIB appears to be associated with the plasma membrane at sites of phagocytosis and was concentrated at the tips of pseudopodia (Baines et al., 1992). This localization suggests that MIB may be involved in membrane dynamics at the cell surface. MIB is regulated by heavy chain phosphorylation of serine 411 (Brzeska et al., 1989, 1990), which is located at the actin-binding site (Rayment et al., 1993). Similar to the myosin-I isoforms in Acanthamoeba, heavy chain phosphorylation results in >20-fold activation of the actin-activated myosin-I ATPase activity (Albanesi et al., 1983). This activation is not the result of changes in the binding of myosin-I to F-actin (Albanesi et al., 1983; Ostap and Pollard, 1996). The transient kinetic studies of Ostap and Pollard (1996) suggest that phosphorylation regulates the rate-limiting phosphate release step, the transition from weakly bound intermediates in rapid equilibrium with actin to strongly bound states, capable of sustaining force.Despite the extensive analysis of ameboid myosin-I biochemical properties and in vivo function, there is little structural information on these myosins. The only detailed structural information available for the myosins-I comes from recent electron microscopy studies on brush border myosin-I (BBM-I) (Jontes et al., 1995; Jontes and Milligan, 1997a ,b; Whittaker and Milligan, 1997), a structurally distinct myosin-I subtype. Therefore, we investigated the structure of a “classical,” ameboid-type myosin, Acanthamoeba MIB using electron microscopy. First, electron micrographs of negatively stained two-dimensional (2D) crystals were used to generate a projection map of MIB at ∼18 Å resolution. In addition, we used cryoelectron microscopy and helical image analysis to produce a moderate resolution three-dimensional (3D) map (30 Å) of actin filaments decorated with MIB. These studies enabled us to compare the structure of MIB with BBM-I. The comparison of MIB with BBM-I reveals marked structural differences in the tail domains of these two proteins; MIB appears to have a much shorter “lever arm” and a more compact tail, whereas most of the BBM-I mass is composed of an extended light chain–binding domain (LCBD). In addition, the MIB catalytic domain appears to be slightly tilted compared to BBM-I, with respect to the F-actin axis. Our structural results suggest that these two types of myosin-I may have distinct intracellular functions.     

C-Nap1, a Novel Centrosomal Coiled-Coil Protein and Candidate Substrate of the Cell Cycle–regulated Protein Kinase Nek2

Nek2 (for NIMA-related kinase 2) is a mammalian cell cycle–regulated kinase structurally related to the mitotic regulator NIMA of Aspergillus nidulans. In human cells, Nek2 associates with centrosomes, and overexpression of active Nek2 has drastic consequences for centrosome structure. Here, we describe the molecular characterization of a novel human centrosomal protein, C-Nap1 (for centrosomal Nek2-associated protein 1), first identified as a Nek2-interacting protein in a yeast two-hybrid screen. Antibodies raised against recombinant C-Nap1 produced strong labeling of centrosomes by immunofluorescence, and immunoelectron microscopy revealed that C-Nap1 is associated specifically with the proximal ends of both mother and daughter centrioles. On Western blots, anti–C-Nap1 antibodies recognized a large protein (>250 kD) that was highly enriched in centrosome preparations. Sequencing of overlapping cDNAs showed that C-Nap1 has a calculated molecular mass of 281 kD and comprises extended domains of predicted coiled-coil structure. Whereas C-Nap1 was concentrated at centrosomes in all interphase cells, immunoreactivity at mitotic spindle poles was strongly diminished. Finally, the COOH-terminal domain of C-Nap1 could readily be phosphorylated by Nek2 in vitro, as well as after coexpression of the two proteins in vivo. Based on these findings, we propose a model implicating both Nek2 and C-Nap1 in the regulation of centriole–centriole cohesion during the cell cycle.The serine/threonine kinase NIMA of Aspergillus nidulans is considered the founding member of a family of protein kinases with a possible role in cell cycle regulation (for reviews see Fry and Nigg, 1995; Lu and Hunter, 1995a ; Osmani and Ye, 1996). In A. nidulans, NIMA clearly cooperates with the Cdc2 protein kinase to promote progression into mitosis (Osmani et al., 1991), and overexpression of NIMA in a variety of heterologous species promotes a premature onset of chromosome condensation (O''Connell et al., 1994; Lu and Hunter, 1995b ). This has been interpreted to suggest evolutionary conservation of a pathway involving NIMA-related kinases (for review see Lu and Hunter, 1995a ). Indeed, kinases structurally related to NIMA are present in many species (Fry and Nigg, 1997). However, the only bona fide functional homologue of NIMA so far isolated stems from another filamentous fungus, Neurospora crassa (Pu et al., 1995), and the functional relationship between vertebrate NIMA-related kinases and fungal NIMA remains uncertain.The closest known mammalian relative to NIMA is a kinase termed Nek2 (for NIMA-related kinase 2)¹ (Fry and Nigg, 1997). This kinase undergoes cell cycle–dependent changes in abundance and activity, reminiscent of NIMA (Schultz et al., 1994; Fry et al., 1995). It is highly expressed in male germ cells (Rhee and Wolgemuth, 1997; Tanaka et al., 1997), and data have been reported consistent with a role for Nek2 in meiotic chromosome condensation (Rhee and Wolgemuth, 1997). However, overexpression of active Nek2 in somatic cells has no obvious effect on chromosome condensation; instead, it induces striking alterations in the structure of the centrosome, the principal microtubule-organizing center of mammalian cells (Fry et al., 1998). Furthermore, immunofluorescence microscopy and subcellular fractionation concur to demonstrate that endogenous Nek2 associates with centrosomes, strongly suggesting that one physiological function of this kinase may relate to the centrosome cycle (Fry et al., 1998).The mammalian centrosome is an organelle of about 1 μm in diameter. It comprises two barrel-shaped centrioles that are made of nine short triplet microtubules and are surrounded by an amorphous matrix known as the pericentriolar material (PCM) (for review see Brinkley, 1985; Vorobjev and Nadehzdina, 1987; Kimble and Kuriyama, 1992; Kalt and Schliwa, 1993; Kellogg et al., 1994; Lange and Gull, 1996). Major progress has recently been made with the demonstration that microtubules are nucleated from γ-tubulin–containing ring complexes (γ-TuRCs), which are concentrated within the PCM (Moritz et al., 1995; Zheng et al., 1995). γ-Tubulin forms complexes with Spc97/98, two evolutionarily conserved proteins first identified in budding yeast spindle pole bodies (Geissler et al., 1996; Knop et al., 1997; Stearns and Winey, 1997), and there is also evidence for an important role of pericentrin and other coiled-coil proteins in organizing γ-TuRCs into higher order lattice structures (Doxsey et al., 1994; Dictenberg et al., 1998). However, in spite of this recent progress, it is clear that the inventory of centrosome components is far from complete.Centrosome structure and function is regulated in a cell cycle–dependent manner (for reviews see Mazia, 1987; Kellogg et al., 1994; Tournier and Bornens, 1994). Once in every cell cycle, and beginning around the G1/S transition, centrioles are duplicated (e.g., Kuriyama and Borisy, 1981a ; Vorobjev and Chentsov, 1982; Kochanski and Borisy, 1990; Chrétien et al., 1997). Late in G2, centrosomes then grow in size (a process referred to as maturation) through the recruitment of additional PCM proteins (Rieder and Borisy, 1982; Kalt and Schliwa, 1993; Lange and Gull, 1995). At the G2/M transition, the duplicated centrosomes separate and migrate to opposite ends of the nucleus. Concomitantly, their microtubule-nucleating activities increase dramatically in preparation for spindle formation (McGill and Brinkley, 1975; Snyder and McIntosh, 1975; Gould and Borisy, 1977; Kuriyama and Borisy, 1981b ; for reviews see Brinkley, 1985; Vorobjev and Nadehzdina, 1987; Karsenti, 1991). By what mechanisms these events are controlled remains largely unknown, but data obtained using phosphoepitope-specific antibodies strongly suggest that phosphorylation of centrosomal proteins plays a major role (Vandré et al., 1984, 1986; Centonze and Borisy, 1990). More direct support for this view stems from the observation that cyclin-dependent kinases (CDKs) enhance the microtubule-nucleation activity of centrosomes at the G2/M transition (Verde et al., 1990, 1992; Buendia et al., 1992) and are involved in promoting centrosome separation (Blangy et al., 1995; Sawin and Mitchison, 1995). Similarly, polo-like kinase 1, a cell cycle regulatory kinase structurally distinct from CDKs, has recently been implicated in centrosome maturation (Lane and Nigg, 1996).The precise role of Nek2 at the centrosome remains to be determined, but it is intriguing that overexpression of this kinase in human cells causes a pronounced splitting of centrosomes. This led us to propose that Nek2-dependent phosphorylation of previously unidentified proteins may cause a loss of centriole–centriole cohesion, and that this event might represent an early step in centrosome separation at the G2/M transition (Fry et al., 1998). With the aim of identifying potential substrates (or regulators) of Nek2, we have now performed a yeast two-hybrid screen, using full-length Nek2 as a bait. We report here the molecular characterization of a novel coiled-coil protein that we call C-Nap1 (for centrosomal Nek2-associated protein 1). C-Nap1 represents a core component of the mammalian centrosome and the first candidate substrate for a member of the NIMA protein kinase family to be identified.     

A Small Conserved Domain in the Yeast Spa2p Is Necessary and Sufficient for Its Polarized Localization

SPA2 encodes a yeast protein that is one of the first proteins to localize to sites of polarized growth, such as the shmoo tip and the incipient bud. The dynamics and requirements for Spa2p localization in living cells are examined using Spa2p green fluorescent protein fusions. Spa2p localizes to one edge of unbudded cells and subsequently is observable in the bud tip. Finally, during cytokinesis Spa2p is present as a ring at the mother–daughter bud neck. The bud emergence mutants bem1 and bem2 and mutants defective in the septins do not affect Spa2p localization to the bud tip. Strikingly, a small domain of Spa2p comprised of 150 amino acids is necessary and sufficient for localization to sites of polarized growth. This localization domain and the amino terminus of Spa2p are essential for its function in mating. Searching the yeast genome database revealed a previously uncharacterized protein which we name, Sph1p (Spa2p homolog), with significant homology to the localization domain and amino terminus of Spa2p. This protein also localizes to sites of polarized growth in budding and mating cells. SPH1, which is similar to SPA2, is required for bipolar budding and plays a role in shmoo formation. Overexpression of either Spa2p or Sph1p can block the localization of either protein fused to green fluorescent protein, suggesting that both Spa2p and Sph1p bind to and are localized by the same component. The identification of a 150–amino acid domain necessary and sufficient for localization of Spa2p to sites of polarized growth and the existence of this domain in another yeast protein Sph1p suggest that the early localization of these proteins may be mediated by a receptor that recognizes this small domain.Polarized cell growth and division are essential cellular processes that play a crucial role in the development of eukaryotic organisms. Cell fate can be determined by cell asymmetry during cell division (Horvitz and Herskowitz, 1992; Cohen and Hyman, 1994; Rhyu and Knoblich, 1995). Consequently, the molecules involved in the generation and maintenance of cell asymmetry are important in the process of cell fate determination. Polarized growth can occur in response to external signals such as growth towards a nutrient (Rodriguez-Boulan and Nelson, 1989; Eaton and Simons, 1995) or hormone (Jackson and Hartwell, 1990a , b ; Segall, 1993; Keynes and Cook, 1995) and in response to internal signals as in Caenorhabditis elegans (Goldstein et al., 1993; Kimble, 1994; Priess, 1994) and Drosophila melanogaster (St Johnston and Nusslein-Volhard, 1992; Anderson, 1995) early development. Saccharomyces cerevisiae undergo polarized growth towards an external cue during mating and to an internal cue during budding. Polarization towards a mating partner (shmoo formation) and towards a new bud site requires a number of proteins (Chenevert, 1994; Chant, 1996; Drubin and Nelson, 1996). Many of these proteins are necessary for both processes and are localized to sites of polarized growth, identified by the insertion of new cell wall material (Tkacz and Lampen, 1972; Farkas et al., 1974; Lew and Reed, 1993) to the shmoo tip, bud tip, and mother–daughter bud neck. In yeast, proteins localized to growth sites include cytoskeletal proteins (Adams and Pringle, 1984; Kilmartin and Adams, 1984; Ford, S.K., and J.R. Pringle. 1986. Yeast. 2:S114; Drubin et al., 1988; Snyder, 1989; Snyder et al., 1991; Amatruda and Cooper, 1992; Lew and Reed, 1993; Waddle et al., 1996), neck filament components (septins) (Byers and Goetsch, 1976; Kim et al., 1991; Ford and Pringle, 1991; Haarer and Pringle, 1987; Longtine et al., 1996), motor proteins (Lillie and Brown, 1994), G-proteins (Ziman, 1993; Yamochi et al., 1994; Qadota et al., 1996), and two membrane proteins (Halme et al., 1996; Roemer et al., 1996; Qadota et al., 1996). Septins, actin, and actin-associated proteins localize early in the cell cycle, before a bud or shmoo tip is recognizable. How this group of proteins is localized to and maintained at sites of cell growth remains unclear.Spa2p is one of the first proteins involved in bud formation to localize to the incipient bud site before a bud is recognizable (Snyder, 1989; Snyder et al., 1991; Chant, 1996). Spa2p has been localized to where a new bud will form at approximately the same time as actin patches concentrate at this region (Snyder et al., 1991). An understanding of how Spa2p localizes to incipient bud sites will shed light on the very early stages of cell polarization. Later in the cell cycle, Spa2p is also found at the mother–daughter bud neck in cells undergoing cytokinesis. Spa2p, a nonessential protein, has been shown to be involved in bud site selection (Snyder, 1989; Zahner et al., 1996), shmoo formation (Gehrung and Snyder, 1990), and mating (Gehrung and Snyder, 1990; Chenevert et al., 1994; Yorihuzi and Ohsumi, 1994; Dorer et al., 1995). Genetic studies also suggest that Spa2p has a role in cytokinesis (Flescher et al., 1993), yet little is known about how this protein is localized to sites of polarized growth.We have used Spa2p green fluorescent protein (GFP)¹ fusions to investigate the early localization of Spa2p to sites of polarized growth in living cells. Our results demonstrate that a small domain of ∼150 amino acids of this large 1,466-residue protein is sufficient for targeting to sites of polarized growth and is necessary for Spa2p function. Furthermore, we have identified and characterized a novel yeast protein, Sph1p, which has homology to both the Spa2p amino terminus and the Spa2p localization domain. Sph1p localizes to similar regions of polarized growth and sph1 mutants have similar phenotypes as spa2 mutants.