Physiol Rev Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Physiol. Rev. 86: 669-707, 2006; doi:10.1152/physrev.00020.2005
0031-9333/06 $18.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (31)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Staub, O.
Right arrow Articles by Rotin, D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Staub, O.
Right arrow Articles by Rotin, D.

Role of Ubiquitylation in Cellular Membrane Transport

Olivier Staub and Daniela Rotin

Department of Pharmacology and Toxicology, University of Lausanne, Lausanne, Switzerland; and Program in Cell Biology, The Hospital for Sick Children, and Department of Biochemistry, University of Toronto, Toronto, Ontario, Canada

ABSTRACT
I. INTRODUCTION
II. UBIQUITYLATION
    A. The Cbl Family of E3 Ligases
    B. Family of HECT Ubiquitin-Protein Ligases
    C. Ubiquitin Binding Domains
        1. UIM motif
        2. Cue domain
        3. UBA domain
        4. UEV domain
        5. GAT domain
        6. NZF domain
        7. Glue domain
III. UBIQUITIN AND INTRACELLULAR MEMBRANE TRANSPORT IN YEAST
    A. Yeast as a Model System to Study Protein Transport
    B. Ubiquitin and Endocytosis in Yeast
        1. Organelles in yeast and mammalian cells involved in protein traffic
        2. The yeast plasma membrane transport proteins
        3. The ABC transport protein Ste6
        4. Yeast permeases
            A) THE GENERAL AMINO ACID PERMEASE, GAP1.
            B) THE URACIL PERMEASE FUR4.
            C) OTHER PERMEFSASES.
        5. The pheromone receptor Ste2
    C. Ubiquitin-Dependent Protein Traffic in Yeast
        1. Role of ubiquitin in Golgi to vacuole transport
            A) PERMEASES.
            B) GGA AND GOLGI TO ENDOSOME TRANSPORT.
        2. Ubiquitin and targeting to the vacuole
            A) SORTING TO THE MVB PATHWAY.
            B) RECEPTORS RECOGNIZING UBIQUITYLATED CARGOES.
            C) PROTEIN COMPLEXES INVOLVED IN TARGETING TO THE MVB PATHWAY.
            D) VPS4.
            E) DOA4.
            F) BRO1.
IV. MAMMALIAN MEMBRANE PROTEINS AND UBIQUITYLATION
    A. Ion Channels
        1. Regulation of ENaC by Nedd4 proteins
        2. Regulation of other ion channels by Nedd4 proteins
        3. Regulation of cell surface CFTR by ubiquitylation
        4. Regulation of other ion channels by the ubiquitin or Sumo systems
    B. Receptor Tyrosine Kinases
        1. Regulation of EGFR endocytosis: a key role for Cbl
    C. T-Cell Receptor Signaling: A Role for Cbl Proteins
    D. Cytokine Receptors
        1. The growth hormone receptor
        2. Other cytokine receptors
    E. The TGF-beta Receptor Pathway
    F. GPCRs
        1. The beta2-adrenergic receptor
        2. Mu and delta opioid receptors
        3. CXCR4 receptor
        4. Other GPCRs
    G. Regulation of Transmembrane Proteins by Ubiquitin in the Nervous System
        1. Ubiquitin and neurological diseases
        2. PD
        3. Regulation of plasma membrane proteins by ubiquitylation in the nervous system
            A) GLUTAMATE RECEPTORS.
            B) NOTCH SIGNALING.
            C) REGULATION OF COMMISSURELESS/ROUNDABOUT IN AXON GUIDANCE.
    H. Viral Budding
VI. PERSPECTIVES
GRANTS
ACKNOWLEDGMENTS
REFERENCES

   ABSTRACT
Top
 Next
 References
 
Ubiquitylation of membrane proteins has gained considerable interest in recent years. It has been recognized as a signal that negatively regulates the cell surface expression of many plasma membrane proteins both in yeast and in mammalian cells. Moreover, it is also involved in endoplasmic reticulum-associated degradation of membrane proteins, and it acts as a sorting signal both in the secretory pathway and in endosomes, where it targets proteins into multivesicular bodies in the lumen of vacuoles/lysosomes. In this review we discuss the progress in understanding these processes, achieved during the past several years.


   I. INTRODUCTION
Top
 Previous
 Next
 References
 
In contrast to soluble cytoplasmic proteins, membrane proteins, which are inserted into a lipid bilayer, cannot freely diffuse throughout the cell and therefore have to be targeted to their destination by specific mechanisms. These mechanisms, generally known as membrane transport or protein traffic between different cellular organelles, follow a common scheme, in which protein cargoes are recruited into budding regions in a donor organelle. These regions then bud off as vesicles and move on to a recipient organelle, with which they fuse and to which they deliver the cargo proteins. These mechanisms have been extensively studied and described over the past two decades (142, 182, 272, 459). One of the dominating themes during the last decade, which has considerably changed our view of protein trafficking, was the discovery that ubiquitylation (i.e., the covalent modification with the polypeptide ubiquitin) is playing a major role in the internalization of membrane proteins, but also during intracellular protein traffic. Especially surprising was the finding that not only membrane proteins themselves become modified by ubiquitin and thereby targeted to their destination, but also the proteins of the trafficking machinery are often themselves ubiquitylated, rendering this machinery highly dynamic. The aim of this review is to provide an overview of these new concepts and use salient examples that highlight the importance of ubiquitin in trafficking, endocytosis, and sorting of transmembrane proteins.


   II. UBIQUITYLATION
Top
 Previous
 Next
 References
 
Ubiquitylation (also referred to as ubiquitination or ubiquitinylation) is a posttranslational modification that involves the covalent attachment of ubiquitin polypeptides to target proteins. Ubiquitin is a highly conserved polypeptide of 76 amino acids; it contains a diglycine motif at its COOH-terminal end, and it is ligated via this COOH terminus to lysine residues of target proteins (Fig. 1A). This modification was originally described as a signal that could target cellular proteins to rapid degradation by a cytosolic complex, the proteasome (160). It turned out to be a highly regulated system that is important for many cellular functions. Ubiquitylation was later found to regulate numerous other processes in the cell (not just serving as a degradation signal for the proteasome), including protein trafficking. For excellent overviews on ubiquitylation, see References 133 325, 345. Here, we will only briefly describe the mechanism leading to ubiquitylation. It involves the action of a cascade of enzymes (Fig. 1, B and C). First, E1 or ubiquitin-activating enzyme forms a thioester bond between its catalytic cysteine and ubiquitin, a process that requires ATP hydrolysis. Only a few E1 enzymes exist (1 in yeast, ~10 in humans) that are involved in ubiquitylation in mammalian cells, whereas at least three other E1 enzymes are involved in the conjugation of ubiquitin-like proteins (163, 367, 402). The ubiquitin moiety is then transferred to an E2, a ubiquitin-conjugating or ubiquitin-carrier enzyme, which also forms a thioester bond between its cysteine and ubiquitin. There are 11 E2 enzymes in yeast and at least 100 in human cells. E2 enzymes then act in concert with E3 enzymes, or ubiquitin-proteins ligases, which are the enzymes that are responsible for substrate recognition. There are hundreds of E3 enzymes (54 in yeast and ~1,000 in humans) (163) comparable to the number of kinases, another indication that ubiquitylation likely plays an important role in cellular regulation, much like phosphorylation. E3 ligases carry out the important tasks of substrate recognition and transfer (or facilitation of transfer) of ubiquitin onto the substrate, usually on lysine residues. There are two major types of E3 ligases: the RING finger E3s and the Hect E3s.


Figure 1
View larger version (47K):
[in this window]
[in a new window]
  		FIG. 1. Schematic view of ubiquitin and the ubiquitylation cascade. A: space filling model of ubiquitin, highlighting its Lys residues and the Ile-44, which has been shown to be involved in binding to ubiquitin binding domains/motifs (PDB code 1UBQ). B and C: ubiquitin is activated by a ubiquitin-activating enzyme, E1, in an ATP-dependent manner that forms a thioester bond between the COOH terminus of ubiquitin and the active site cysteine. Ubiquitin is then transferred to the active site cysteine of a ubiquitin-conjugating enzyme, E2. B: RING finger E3 ubiquitin-protein ligases promote the transfer of ubiquitin from the E2 onto the lysine of a target protein, thereby forming an isopeptide bond. C: HECT domain containing E3 ligases form a thioester bond via their catalytic cysteine and then carry out isopeptide formation between ubiquitin and the substrate. E4 enzymes can stimulate the formation of polyubiquitin chains, and deubiquitylating enzymes (DUBs) reverse the ubiquitylation reaction.

 
The RING (really interesting new gene) domain-containing E3s bring the E2 enzyme in close proximity of the target protein, allowing the E2 to directly ubiquitylate the substrate. (428). The RING domain is defined by the concensus CX2CX(9–39)CX(1–3)HX(2–3)C/HX2CX(4–48)CX2C, in which the cysteines and histidines bind Zn2+ (428). This arrangement produces a structure that is referred to as the "cross-brace motif." This structure is stabilized by the Zn2+ coordination for the first Zn2+ atom by cysteines in positions 1, 2, 5, and 6, and for the second Zn2+ atom by cysteines or histidines in positions 3, 4, 7, and 8. The RING fingers are classified into RING-HC and RING-H2 motifs depending on whether a cysteine or histidine occupies the fifth coordination site, respectively. Structures of RING-HC fingers show two interleaved Zn2+ binding sites, which is different from the tandem arrangement of Zn2+ binding sites characteristic of Zn2+ fingers (119). An example of a RING-containing E3 is Cbl (Fig. 2A), which plays an important role in the ubiquitylation of plasma membrane proteins.


Figure 2
View larger version (28K):
[in this window]
[in a new window]
  		FIG. 2. Cbl and Nedd4/Nedd-like families. A: Cbl family: human, Drosophila, (D-Cbl) and C. elegans (SLI-1) Cbl family members are shown. The Cbl proteins are highly conserved in the NH2-terminal region where they comprise a tyrosine-kinase binding domain (TKB), which is composed of a 4-helix bundle (4H), a calcium binding EF domain, and a variant SH2 domain that is linked via a linker with the RING finger domain. In the COOH-terminal half are a number of proline-rich regions (indicated with orange boxes), a ubiquitin-associated UBA domain, and a leucine zipper. v-Cbl is a retroviral oncogenic form of Cbl lacking the RING finger domain and the rest of the COOH terminus. B: Nedd4/Nedd4-like family members from yeast, worm, fly, and mammalian species are shown. They all contain a C2 domain (which in some cases can be spliced out, dashed lines), WW domains, and a HECT domain. Diagrams are not to scale.

 
The second class of E3 enzymes involves the HECT (homologous to E6-AP COOH terminal) domain containing ubiquitin-protein ligases, which also directly interact with E2 enzymes (173). However, here ubiquitin is first transferred onto a conserved cysteine in the HECT domain, and it is the HECT E3 enzyme that ubiquitylates the substrate protein. The HECT containing E3 enzymes includes the family of Nedd4 (neuronal precursor cell expressed developmentally downregulated)/Nedd4-like ubiquitin-protein ligases (Fig. 2B), many of which are involved in the ubiquitylation of membrane proteins.

Ubiquitin is usually ligated onto lysine residues of the substrate or of ubiquitin itself, forming di- or polyubiquitin chains in the latter. Hence, proteins can either be monoubiquitylated (one ubiquitin polypeptide on a single lysine), multiubiquitylated (several lysines modified with just one ubiquitin), diubiquitylated (diubiquitin on substrate lysine), or polyubiquitylated (extended polyubiquitin chain). While cytosolic proteins destined for destruction by the proteasome are polyubiquitylated, during trafficking, membrane proteins and proteins of the trafficking machinery are generally monoubiquitylated (383) or diubiquitylated (124). There are seven lysines in ubiquitin itself (Fig. 1A). Although Lys-48 and Lys-63 are the most frequently utilized residues to form polyubiquitin chains, linkage through the other five lysines has been reported (319). The mechanisms that lead to polyubiquitylation are not very well understood. A novel class of enzymes, the E4 enzymes containing a U-box motif, have been proposed to be involved in this process, but ubiquitin-protein ligases may be involved in the process as well (158, 223, 243), and U-box containing proteins can function also as E3 enzymes (189). Modification with ubiquitin is a reversible process. This is achieved by the action of isopeptidases, or deubiquitylating enzymes. Although there are probably hundreds of such enzymes, deubiquitylation has so far not been extensively studied, but it is likely that this is as important as ubiquitylation itself and also highly regulated (8, 217). There are a number of small polypeptides with homology to ubiquitin, commonly known as ubiquitin-like proteins. Examples include SUMO, Nedd8, ISG15, and FAT10. These ubiquitin-like proteins are targeted to substrate proteins using enzymatic cascades highly similar, but not identical, to the ubiquitylation cascades, including E1, E2, E3, and deubiquitylation enzymes (for a review, see Refs. 367, 402). The following is a brief overview of E3 ubiquitin-protein ligase families particularly pertinent to this review.

A. The Cbl Family of E3 Ligases

Cbl (Casitas B-lineage lymphoma) was first discovered as a viral oncogene in 1989 in a study involving a retrovirus that induces early B-cell lymphomas (234) and was called v-Cbl. Cloning of its murine cellular homolog revealed that v-Cbl encoded only the first 355 amino acids of the murine protein, c-Cbl (39). Two other members of the Cbl family have been identified in mammals, Cbl-b and Cbl-3 (210, 211, 423) (Fig. 2A). These proteins share a tyrosine kinase binding (TKB) domain that includes a variant SH2 domain, a linker, a RING finger domain, and in c-Cbl and Cbl-b, also a COOH-terminal region that contains a proline-rich region and a ubiquitin association (UBA) domain. The oncogenic v-Cbl comprises only the NH2 terminus and lacks the linker region and the RING finger domain (422). It has been shown that the linker region interacts with the RING finger domain, permitting an optimal placement of an E2 enzyme to allow transfer of ubiquitin onto the substrate protein (485). v-Cbl, which lacks the RING finger, is not able to ubiquitylate substrates and acts in a competitive fashion towards c-Cbl. For a detailed review on Cbl proteins, see Reference 422.

B. Family of HECT Ubiquitin-Protein Ligases

This family is part of the HECT domain containing ubiquitin-protein ligases (Fig. 2B). Its best studied member, Nedd4, was identified in a subtractive screen between neuronal precursor cells and adult brain cells (232). There are nine members encoded in the human genome, and all are characterized by the presence of an NH2-terminal C2 (calcium-dependent lipid binding domain) (for a review, see Ref. 351), two to four WW (protein-protein interaction) domains (reviewed in Refs. 354, 405), and a COOH-terminal HECT domain (173) (Fig. 2B). The C2 domain and the WW domains are involved in localization and substrate recognition, respectively (197, 200, 326, 327), whereas the HECT domain is an E3 ubiquitin-protein ligase domain. WW domains of Nedd4 proteins bind a short recognition motif called the PY motif (L/PPxY) (200, 202). Nedd4/Nedd4-like proteins are involved in a plethora of functions, including ubiquitylation of membrane proteins or of proteins of the trafficking machinery. Reviews on Nedd4/Nedd4-like proteins can be found in References 154, 179, 354.

C. Ubiquitin Binding Domains

There is a growing list of ubiquitin binding domains and motifs ("ubiquitin receptors") (Fig. 3). These define the cellular proteins that recognize and interact with ubiquitin and translate ubiquitin signal into a cellular action, such as degradation by the proteasome, internalization involving Eps15, or targeting to the multivesicular body.


Figure 3
View larger version (13K):
[in this window]
[in a new window]
  		FIG. 3. Examples of ubiquitin binding motifs/domains (in color) present in proteins regulating trafficking, endocytosis, and sorting of cellular cargo. Only key domains/motifs (in addition to the ubiquitin binding domains) are shown. Figure is not to scale. CB, clathrin binding motif. See text for other definitions.

 

1. UIM motif

The UIM (ubiquitin-interacting motif) was first described as a peptide sequence consisting of a highly conserved Ac-Ac-Ac-Ac-{Phi}-X-X-Ala-X-X-X-Ser-X-X-Ac, where {Phi} represents a large hydrophobic residue (typically Leu) and Ac represents an acidic residue (Glu, Asp) (115, 168). This motif was originally found in the ubiquitin-binding region of the Rpn11 proteasome subunit and shown to bind to polyubiquitin (121, 482). The motif was subsequently identified in many other proteins (50, 168, 328), often in combination with other protein-protein interaction motifs. Many of these proteins are either part of the ubiquitin system or play a role in protein trafficking (see below). These also include two different deubiquitylating enzymes [KIAA1594 (USP37) and USP25], an F-box protein (Ufo1), a HECT-type ubiquitin-protein ligase from plants (Upl1), but no E1 or E2 enzyme. UIM containing proteins involved in trafficking include the following (Fig. 3): 1) the Eps15 subfamily of EH proteins (219, 329), 2) the Ent1/2-epsin subfamily of ENTH proteins (381), 3) the FYVE domain proteins including Vps27/Hrs (381), and 4) the SH3-containing proteins STAM and HBP (168). Eps15 and its ortholog Ede1 in Saccharomyces cerevisiae comprise two to three EH (Epsin15 homology) domains (110), which bind preferentially to the NPF motifs of epsin/Ent1/2 (4, 316, 329, 359). A number of proteins containing UIM motifs are ubiquitylated, including Hrs/Hgs (205, 265), Eps15 (219), epsins (308, 329), and likely STAM. Often, the UIM domain is required for the ubiquitylation of the protein that harbors it and appears to recruit the relevant ubiquitin-protein ligase (205, 219, 308, 329). The significance of this dual presence of ubiquitin moiety and UIM motif on the same proteins is not clear but may facilitate the formation of complexes with other proteins that contain ubiquitin-binding motifs/domains. The UIM motif by itself does not contain any lysine residues and hence is not directly modified by ubiquitin. It may recruit ubiquitin-protein ligases, for example, via ubiquitin linked by thioester to HECT-type ubiquitin-protein ligases. After transfer of the ubiquitin onto the protein, the HECT E3 enzyme is released. Alternatively, an E3 enzyme may be recruited by the UIM motif, promoting ubiquitylation. The presence of the UIM motif and monoubiquitin on the same protein may also lead to intra- or intermolecular interactions. Intramolecular interactions may hide internal lysines on ubiquitin and thereby interfere with polyubiquitylation of the protein. As ubiquitylation usually is a regulated process and also reversible, this likely suggests that there are dynamic networks of interacting proteins that are controlled by the ubiquitin system (88).


2. Cue domain

The Cue domain was originally identified by database searches and proposed to be scaffolding domain for ubiquitylated proteins (27, 331). It is present in the Cue protein, which binds ubiquitin; is associated with the endoplasmic reticulum (ER) membrane; and recruits the E2 enzyme Ubc7 to the ER membrane (34). The Cue domain is also present in Vps9, a guanidine nucleotide exchange factor involved in yeast endocytosis (73, 91, 335, 382); in Tollip, an intermediate in interleukin-1 signaling; in AMFR, a cytokine receptor involved in tumor cell motility and metastasis; and in AUP1, a ubiquitous protein involved in integrin signaling (331). Like the UIM motif, Cue binds to both monoubiquitin and polyubiquitin and promotes ubiquitylation of Vps9 (73, 201, 382). The structure of a Cue domain associated with monoubiquitin has been solved (201, 335). The contact sites between the Cue domain and ubiquitin are conserved hydrophobic residues in the latter, including Leu-8, Ile-44, and Val-70. These same amino acids were shown previously to be important for receptor endocytosis (30, 389), and in vitro binding assays illustrated a requirement of this hydrophobic patch for interaction with both the UIM and Cue domains (381, 382). The crystal structure of the Vps9-Cue domain bound to monoubiquitin also shows that the contact surface covers Lys-48, which may represent an occlusion mechanism for inhibiting polyubiquitin formation (201). The cue domain shares architecture very similar to the UBA domain (see below) (307), which also has a hydrophobic contact points with ubiquitin. It is likely that the Cue domain and the UBA domain bind ubiquitin in a similar fashion.


3. UBA domain

The UBA (ubiquitin associated) domain was also originally identified by bioinformatics means in proteins that were known to interact with ubiquitin (167). It is present in many proteins involved in the ubiquitylation cascade such as E2, E3, and deubiquitylation enzymes. Of interest for this review is c-Cbl, the E3 ligase involved in ubiquitin-dependent downregulation of many receptor tyrosine kinases, as well as Ede1p (the yeast homolog of Eps15) (340), which contain a UBA domain. The three-dimensional structure of the UBA domain has been solved and shown to be a compact three-helix bundle and to bind to ubiquitin in a very similar manner as does the Cue domain (87, 307, 463). It is likely able to replace functionally the UIM domain, as mammalian Eps15 contains two UIM domains at the COOH terminus, while the yeast homolog has a UBA domain instead.


4. UEV domain

The UEV (ubiquitin E2 variant) domain is homologous to the catalytic UBC (ubiquitin-conjugating) domain of E2 enzymes involved in the conjugation of ubiquitin to target proteins, but it does not contain the catalytic cysteine essential for E2 activity (226, 332). This domain is found in a number of different proteins. Of interest, it is included in the NH2 terminus of Tsg101 (Vps23), one of the proteins involved in the multivesicular body (MVB) pathway, i.e., for sorting ubiquitylated cargo to the lumen of the lysosomes (see below). In this context, Tsg101-UEV:ubiquitin interactions have been shown to be essential for the MVB pathway (36, 206, 255). The structure of the Tsg101-UEV domain complexed to ubiquitin has been solved (410). The UEV domain has essentially the same structure as in the unbound form, which resembles the canonical E2 enzymes and adopts the typical {alpha}/beta-fold, but has an additional NH2-terminal helix and lacks two COOH-terminal helices (333).


5. GAT domain

The GAT (GGA and Tom1 homologous) domain is a ubiquitin binding module that plays a role in protein trafficking. It was originally described as a domain present in GGA (Golgi localized {gamma} ear containing ARF binding) proteins (see below)(81), and later described in a number of proteins shown to participate in intracellular trafficking and sorting. The GAT domain is composed of an all {alpha}-helices fold that comprises two subdomains. An NH2-terminal "hook" that can bind ARF, and a COOH-terminal triple {alpha}-helix bundle, that can interact with Rabaptin 5 and ubiquitin (67, 336, 369, 378, 379, 409, 469, 486, 487) (see below).


6. NZF domain

The NZF domain is another domain that binds ubiquitin and is present on proteins that are involved in intracellular protein trafficking. Its name is derived from NP14 zinc finger (or RANBP2/Nup358 zinc finger) (17, 80, 280, 296). NP14 NZF forms a compact module composed of four anti-paralleled beta-strands linked by three ordered loops. A single Zn2+ is coordinated by four conserved cysteines from the first and the third loop, which form two rubredoxin knuckles. It binds weakly to free ubiquitin, using a conserved threonine-phenylalanine dipeptide that interacts with Ile-44 (5, 448). The NZF domain has been identified in Vps36, which plays a role in sorting cargoes into the MVB pathway (see below) (5, 17). However, the NZF domain is not present in mammalian Vps36, where ubiquitin binding appears to be achieved via a novel domain, the Glue domain.


7. Glue domain

The Glue (Gram-like ubiquitin binding in Eap45) domain was identified in Eap45, the mammalian ortholog of Vps36. As is the case for NZF in yeast Vps36, this domain is situated at the NH2 terminus of Eap45. It binds to ubiquitin in the micromolar range, comparable to the other ubiquitin binding domains. The GLUE domain shares similarities in its primary and predicted secondary structures to phosphoinositide-binding GRAM and PH domains. Accordingly, Eap45 binds to a subset of 3-phosphoinosites, suggesting that ubiquitin recognition could be coordinated with phosphoinositide binding (388).

Interestingly, while the various ubiquitin binding domains described above adopt different folds/conformation from each other, they all appear to bind the same hydrophobic surface patch of ubiquitin that includes I44 (Fig. 1A) (163).


   III. UBIQUITIN AND INTRACELLULAR MEMBRANE TRANSPORT IN YEAST
Top
 Previous
 Next
 References
 
A. Yeast as a Model System to Study Protein Transport

For more than two decades, the yeast cell model played a leading role in the elucidation of molecular mechanisms involved in protein trafficking, initiated by the classical studies of Schekman and collaborators (303), involving the screening for a large numbers of secretion (Sec) mutants. Studying trafficking in yeast has been very useful due to the powerful combination of genetics with biochemistry and cell biology afforded in this eukaryote (207), and the conservation of many of the relevant pathways in higher organisms. Ubiquitylation of membrane proteins plays a role at different stages of their lifespan: 1) during and after biosynthesis in the ER, when quality control and ER-associated degradation (ERAD) take place; 2) in the trans-Golgi Network (TGN) for sorting of proteins to different cellular destinations; 3) at the plasma membrane for internalization of plasma membrane proteins; and 4) sorting and translocation to MVBs in the late endosome. Two classes of yeast plasma membrane proteins have been extensively studied with respect to ubiquitylation: permeases/transport proteins and G protein-coupled seven-transmembrane receptor (GPCR). The following section summarizes the knowledge on the involved pathways, including the implicated organelles, and the role of ubiquitylation of the above mentioned classes of proteins.

B. Ubiquitin and Endocytosis in Yeast


1. Organelles in yeast and mammalian cells involved in protein traffic

Protein trafficking in the cell refers to protein transport between various organelles or subcellular compartments. Most of these organelles (i.e., nuclei, ER, Golgi, lysosomes, mitochondria, endosomes, peroxisomes) are well characterized. However, it is less appreciated that the organelles involved in the endocytic and lysosomal pathway are highly dynamic (reviewed by Maxfield and McGraw, Ref. 272). Endocytosis (the internalization of membrane proteins and soluble extracellular proteins) is achieved by various mechanisms, including clathrin- and non-clathrin-mediated endocytosis, caveolae formation, and pinocytosis. The best studied is clathrin-coated endocytosis (291). Because of its dynamic nature, it has been difficult to clearly define organelles within the endocytic pathway. An "organelle," such as the "sorting endosome," often matures into another type of organelle. "Resident" proteins of an organelle are constantly escaping from their organelle and are then retrieved by specific retrieval mechanisms. Hence, it is difficult to define such an organelle just by the presence of certain marker proteins, or by morphological features. It is therefore necessary to use several characteristics (composition, morphology, and function) to define them.

The organelles of the endocytotic pathway have been originally divided into early and late endosomes, whereby early endosomes are defined as the organelles that can be directly reached by vesicles derived from the plasma membrane, and from which direct recycling to the plasma membrane still occurs (272), whereas such processes are no longer possible from the late endosome. Early endosomes can be subdivided into sorting endosomes and endocytotic recycling compartment (ERC). The sorting endosomes are formed by internalized vesicles. They have a half-life of ~5–10 min. Proteins in the sorting endosome can travel to three different destinations: 1) to the plasma membrane, which is the default pathway for transmembrane proteins; 2) to the ERC; and 3) to the late endosome (by maturing of the organelle, and consequently the default pathway for soluble proteins and specific targeting pathway for transmembrane proteins). This specific targeting of membrane proteins to the late endosome is achieved in many cases through their ubiquitylation.

The late endosome can be distinguished morphologically from sorting endosomes by their more spherical appearance compared with the more tubular sorting endosomes. Late endosomes tend also to be located closer to the nucleus. Moreover, parts of the late endosomes have a multivesicular appearance. The elegant cell biological and genetic studies that led to the characterization of the multivesicular (MVB) pathway are described below. Late endosomes containing MVBs fuse then with their outer, limiting membrane with the lysosome (or vacuole in yeast), thereby releasing the MVBs into the lysosomal lumen, where they are dissociated and their cargoes degraded (207).


2. The yeast plasma membrane transport proteins

This group of plasma membrane proteins includes two proteins of the ABC transporter family, namely, Ste6, which secretes the yeast mating pheromone a-factor, and Pdr5, involved in multiple drug resistance in yeast (104), as well as the family of transporters of the MFS (multi-facilitators superfamily; often referred to as permeases), i.e., proteins that have a central hydrophobic core of 10–12 transmembrane domains, flanked by cytoplasmic NH2 and COOH termini (13, 127). Two permeases of this family have been particularly well studied: the general amino acid permease Gap1 and the uracil permease Fur4. A mammalian member of the ABC transporter family, cystic fibrosis transmembrane conductance regulator (CFTR), has also been extensively studied in yeast, especially with respect to ER-associated degradation.


3. The ABC transport protein Ste6

The first evidence of a role for ubiquitin in the control of cell surface expression of a plasma membrane protein came from work on Ste6, an ABC transport protein required for secretion of the mating pheromone a-factor (231, 277). In this original work it was shown that Ste6 accumulates in endocytosis defective mutants as a ubiquitylated protein, suggesting that ubiquitylation of Ste6 occurs on the way to or at the plasma membrane (224). In support of a role of the ubiquitin system in the regulation of Ste6 was the finding that its rapid turnover was attenuated in ubc4ubc5 mutants. Ubiquitylation of Ste6 is mediated by the region linking the two halves of the molecule, referred to as D-box (destabilization box). The D-box is composed of an upstream region containing mostly acidic residues (A-box) and a downstream region with basic residues (B-box). The A-box contains a motif, "DAKTI," which closely resembles the DAKSS motif in the Ste2 receptor (225, 352) (see below). Deletion of the A-box leads to the accumulation of Ste6 at the plasma membrane, whereas wild-type Ste6 is mostly intracellular. A deubiquitylating enzyme, Ubp1, plays a role either in internalization or in the recycling of Ste6 to the plasma membrane, as its overexpression causes accumulation of Ste6 at the cell surface. However, Ste6 does not seem to be the direct target of Ubp1, since its ubiquitylation level is not affected by Ubp1 (364). Ste6 targeting into the lumen of the late endosome or vacuole (MVB pathway) for degradation requires the action of Doa4, another deubiquitylating enzyme also known as Npi2/Ubp4/Ssv7, as Ste6 accumulates in the vacuole in doa4 mutants (250, 253). However, it is known that Doa4 is required for efficient ubiquitin homeostasis, because free ubiquitin levels are reduced significantly in doa4 mutants (397, 411). Indeed, overexpression of ubiquitin restores targeting of Ste6 to the MVB pathway and accompanied degradation, suggesting that lack of free ubiquitin was the cause of the impaired MVB targeting (253). It is not yet clear which degradation systems are involved in Ste6 turnover, as its turnover is not only slowed down in the vacuolar protease mutant pep4 (224, 251), but also in mutants defective in the chymotrypsin-like activity of the proteasome (pre1–1/pre2–1 double mutant) (251). Whether this means a cooperate activity of vacuolar proteases and the proteasome at the level of the vacuole remains to be determined. Alternatively, the observed impairment of Ste6 turnover may be due to degradation of misfolded Ste6 proteins in the ER, which requires proteasome activity. Another protein of the ABC transport family in yeast is Pdr5, which has also been shown to be regulated by ubiquitylation (103).


4. Yeast permeases

Remarkably, the expression, activity, and turnover of yeast permeases is highly regulated and allows a yeast cell to adapt to rapidly changing extracellular milieu conditions and nutritional availability (171). For most of these proteins the regulation controls the steady-state expression at the cell surface. Indeed, there has been only one case so far of a yeast plasma membrane protein whose activity is not controlled by protein trafficking, namely, the copper-transport protein (Ctr1), a three-transmembrane domain spanning protein which undergoes copper-induced proteolysis at the plasma membrane (310). Most of the other permeases are regulated at least in part by the ubiquitin system (43, 99, 161, 354). The emphasis of this review focuses on recently developed concepts with respect to molecular mechanisms involved in ubiquitylation-dependent membrane protein transport, using the Gap1 and Fur4 permeases as examples.

A) THE GENERAL AMINO ACID PERMEASE, GAP1.  Evidence that ubiquitylation may be involved in the control of internalization of permeases came originally from genetic studies. It had been known for a long time that yeast cells control precisely the uptake of organic nutrients. This control is exerted by a process referred to as "catabolite inactivation," during which sugar or amino acid transporters are rapidly downregulated. This is achieved by the addition of either glucose, the favored carbon source of yeast, or by the addition of NH4+, which is preferred as a nitrogen source over amino acids. The latter, nitrogen catabolite inactivation, had been studied for years and led to the identification of yeast mutants that were deficient in the inactivation of amino acid uptake (referred to as npi1, npi2, and npi3) (140, 141, 460). In this context it was shown that Gap1 was affected by these mutants and was not downregulated. Molecular characterization of these mutants showed that the defects were in enzymes involved in ubiquitylation, namely, npi1 in the gene encoding the HECT domain containing ubiquitin-protein ligase Rsp5 (Fig. 2B) (155) and npi2 in the gene encoding the deubiquitylation enzyme Doa4 (397). Indeed, NH4+ stimulates ubiquitylation and endocytosis of Gap1 and subsequent degradation in the vacuole. In npi1 mutants (expressing 10% of Rsp5), ubiquitylation is impaired, Gap1 is not internalized, and its turnover is attenuated (155, 395). Moreover, ubiquitylation and downregulation of Gap1 is dependent on enzymatic activity of Rsp5 (396). The same authors also found that NH4+ stimulates the modification of Gap1 via lysine-63 linked ubiquitin chains, although blocking the formation of K63-linked chain does not completely inhibit NH4+-induced internalization of Gap1 (397).

Analysis of the second npi mutant, npi2, bearing a mutation in a gene encoding Doa4 revealed that Gap1 is regulated by deubiquitylation (397). Similar to Ste6, overexpression of ubiquitin in npi2 cells restored free ubiquitin levels and proper ubiquitylation, internalization, and degradation of Gap1, suggesting that Doa4 regulates Gap1 indirectly via impairing the generation of free ubiquitin levels. Gap1 is also regulated by another deubiquitylating enzyme, Dot4, which appears to stimulate Gap1 activity, as suggested by its loss of function in {Delta}Dot4 cells (195). Whether Dot4 is acting via direct deubiquitylation of Gap1 or via another substrate remains to be demonstrated.

A third npi mutant (npi3) was identified in nitrogen-induced catabolite inactivation. NPI3 (corresponding to the BRO1 gene, see below) is required for NH4+-induced downregulation and efficient ubiquitylation of Gap1 (398) and appears to be important for efficient sorting of Gap1 into the vacuole (see below).

Sorting of Gap1 is also controlled by the Npr1 kinase (433), which stabilizes Gap1 at the plasma membrane. Inactivation of Npr1 activity causes direct targeting to the vacuolar pathway, rapid internalization, and increased ubiquitylation of Gap1, suggesting that Npr1 counteracts the ubiquitylation of Gap1 by Rsp5 (79).

B) THE URACIL PERMEASE FUR4.  The uracil permease Fur4 was the second permease for which an Npi1/Rsp5-dependent inactivation was demonstrated (155). Although this permease undergoes constitutive endocytosis, this endocytosis is increased under stress conditions, such as elevated temperature, nutrient starvation, and inhibition of protein translation (125, 444), or in the presence of excess of uracil (372). As a result, Fur4 enters the endocytic pathway and is degraded in the vacuole. Like Gap1, Fur4 becomes ubiquitylated in a Rsp5-dependent manner, and the ubiquitylated species accumulate at the plasma membrane in cells deficient in the internalization step of endocytosis (125, 286). In npi1 mutant cells, Fur4 accumulates at the nonpermissive temperature at the plasma membrane, and the protein is stabilized (125). This suggests that ubiquitylation is necessary for internalization of the permease. Moreover, Fur4 is not degraded by the proteasome but is stabilized in cells deficient in vacuolar proteases (125). Further evidence for a role of ubiquitylation in the internalization of Fur4 comes from experiments showing that cells lacking Doa4 are also defective in uracil permease ubiquitylation and internalization. This phenotype can be corrected by overproducing ubiquitin, suggesting that inactivation of Doa4 leads to a lack of free ubiquitin available for ubiquitylation (124). This system allows studying the type of ubiquitylation that is important for Fur4 endocytosis. Accordingly, overexpression of ubiquitin mutants that carry Lys to Arg mutations at Lys-29 and Lys-48 restore Fur4 ubiquitylation, whereas a ubiquitin mutated at Lys-63 does not, suggesting that Fur4 is ubiquitylated by ubiquitin chains extended through Lys-63. Similar to Gap1, when polyubiquitylation is blocked, the Fur4 permease still undergoes endocytosis, but at a reduced rate. This suggests that monoubiquitylation is sufficient to induce permease endocytosis, but that Lys-63-linked ubiquitin chains appear to augment this process (124).

C) OTHER PERMEFSASES.  The maltose transporter is another permease whose internalization is induced by nitrogen starvation if a fermentable carbon source is present and for which this internalization depends on Rsp5 and Doa4 (256, 278). The impaired internalization in doa4 mutant can be restored by overexpression of ubiquitin, indicating that it is mostly the low levels of ubiquitin that interfere with internalization of this transporter (257). Moreover, monoubiquitylation appears to be sufficient as an internalization signal.

The tryptophan permease Tat2 is internalized in a manner dependent on high pressure and requiring the activity of Rsp5, the Rsp5-associated proteins Bul1 and Bul2 (1), and the vacuolar sorting protein Vps27 (295). Similar to Fur4 and Ste6, the internalization and degradation of Tat2 also depends on Doa4; however, in this case, two other deubiquitylating enzymes, Ubp6 and Ubp14, are also required (283). Overexpression of ubiquitin does not rescue the defect of any of these DUBs, indicating that they are not only involved in ubiquitin recycling, but also have other specific effects. Ubp6 is a protein that is associated with the base of the regulatory unit of the proteasome and may be involved in the removal of ubiquitin from substrates, and appears to be important for the recycling of ubiquitin (9, 239). Ubp14 is involved in the disassembly of free polyubiquitin chains, which correlates with ubiquitin-dependent defects of the proteasome (7). It is therefore likely that these enzymes play a specific role in high-pressure-dependent internalization and degradation. Ubiquitylated Tat2p may be recognized by the endosomal machinery involved in translocation into the MVBs, in the course of which Doa4 may play a role in deubiquitylation of the enzyme. Ubp6 and Ubp14 may play a role here as well, in conjunction with the proteasome. It will be necessary to establish if these deubiquitylating enzymes are directly acting on Tat2p (283).


5. The pheromone receptor Ste2

The {alpha}-factor receptor Ste2 is a GPCR that binds the mating pheromone {alpha}-factor. Upon ligand binding it activates a signal-transduction pathway and undergoes rapid internalization and degradation in the vacuole. {alpha}-Factor-dependent stimulation triggers the rapid appearance of ubiquitylated Ste2 species, which accumulate at the plasma membrane in endocytosis-deficient end4{Delta} cells (162). The ubiquitylation and accompanied endocytosis of Ste2 depends on the E2 enzymes Ubc1, Ubc4, and Ubc5 and on the E3 enzyme Rsp5 (95, 162). Ste2, carrying a deletion of two-thirds of its cytoplasmic COOH-terminal tail, still internalizes in an {alpha}-factor-dependent manner (352). This truncated receptor contains a short sequence of nine amino acids, SINNDAK337SS, which plays a critical role in internalization. K337, within the SINNDAKS motif, is essential for ubiquitylation of the truncated receptor, as are the three serines in this sequence (162). Phosphorylation of the three serines in response to receptor stimulation is a prerequisite for receptor ubiquitylation (164). This is similar to the phosphorylation of the PEST sequence in Fur4 that is required for its ubiquitylation (263). Moreover, as is the case for Fur4, it appears that casein kinase I is the kinase that phosphorylates the receptor (164, 262). Monoubiquitylation was found to be sufficient for internalization of Ste2, since overexpression of ubiquitin with all its lysines mutated to arginines is sufficient to efficiently internalize Ste2 upon {alpha}-factor stimulation, as was fusion of a lysine-less ubiquitin moiety to the (lysine-less) cytoplasmic tail of Ste2 (421). With the use of this model it was demonstrated that a hydrophobic patch on the surface of ubiquitin, comprising Phe-4 and particularly Ile-44, serves as signal for internalization (383, 389).

Ste2 has been used as a model for studying the molecular mechanisms of internalization of ubiquitylated proteins. It has been demonstrated that proteins involved in endocytosis contain ubiquitin binding domains or motifs, such as the UIM, UBA, or CUE motifs/domains described above. When systematically mutating proteins containing such domains, several of them were found to display endocytosis defects (381). Yeast strains that were mutated in EDE1, VPS27, ENT1, and ENT2 had endocytosis defects. Ent1 and Ent2 have UIM domains and are homologs of the mammalian epsins, required for the internalization step of endocytosis (59). Ede1 is a UBA-containing protein that functions in fluid-phase endocytosis and receptor-mediated endocytosis. It is the homolog of mammalian Eps15, which contains a UIM motif, and is a component of the mammalian endocytosis machinery (31). All these proteins bind monoubiquitin in vitro. The binding depends on the UIM motif (Vps27), involves Ile-44 of ubiquitin, and in vitro assays show that UIM of Ent1 and Vps27 bind directly to ubiquitin (381). Ede1, Ent1, and Ent2 play a role in internalization of {alpha}-factor. In ent1{Delta}ent2{Delta} cells, carrying an allele with a temperature-sensitive mutation in ent1p (ent1ts), the internalization of {alpha}-factor is severely impaired at the nonpermissive temperature. Cells expressing ent1{Delta}UIM have an internalization defect that is milder compared with ent1ts cells. In mammalian cells, epsins and Eps15 have been shown to interact via a NPF motif and the EH (epsin homology) domain (60). These domains are conserved in yeast, making it likely that the complex is conserved as well. It was therefore speculated that the Ede1 UBA domain interacts with ubiquitin and that this may be sufficient to promote internalization. Indeed, ent1{Delta}ent2{Delta}ede1{Delta} cells carrying ent1{Delta}UIM are defective in {alpha}-factor internalization, indicating that Ede1 overcomes the defect of ent1{Delta}UIM.

Internalization of the {alpha}-factor receptor and fluid-phase endocytosis depends also on the yeast proteins Rsv167 and Sla1 (28, 292). Rsv167 is the yeast homolog of endophilin/amphiphysin, and Sla1 of CIN85 (Cbl interacting protein 85) (416), proteins that are both known to be involved in mammalian receptor internalization (see below). Both Rsv167 and Sla1 bind to the ubiquitin-protein ligase Rsp5, and Rsv167 becomes monoubiquitylated on Lys-481, which is situated in the COOH-terminal SH3 domain of Rsv167 (400). Monoubiquitylation within the SH3 domain may control protein-protein interaction but is not required for endocytosis or general function of Rsv167.

C. Ubiquitin-Dependent Protein Traffic in Yeast

In recent years it has become evident that ubiquitin plays also an essential role in intracellular protein trafficking, in protein targeting either from the Golgi or from the plasma membrane via the endosomal system to the MVBs in the vacuolar lumen. Again, much of the current understanding was first elucidated in yeast and subsequently in mammalian cells.


1. Role of ubiquitin in Golgi to vacuole transport

A) PERMEASES.  Ubiquitylation is not only important for the internalization of plasma membrane proteins, but it also plays a role in their sorting. Helliwell et al. (158) showed that the general amino acid permease Gap1 becomes either monoubiquitylated or polyubiquitylated and that both types of modifications depend on Rsp5. Two Rsp5-associated proteins, Bul1 and Bul2, are required for polyubiquitylation (158). Bul1 and Bul2 associate via their PY motifs with the WW domains of Rsp5 (472, 473) and promote polyubiquitylation of Gap1, thereby targeting Gap1 into the vacuolar pathway. When Bul1p and Bul2p are overexpressed, Gap1 is sorted to the vacuole independently of the nitrogen source, whereas deletion of these proteins leads to the constitutive expression at the cell surface and diminution of polyubiquitylation of Gap1. These data therefore suggest that Bul1p/Bul2p can act as an E4 enzyme, promoting polyubiquitylation of Gap1, and consequently targeting it to the vacuole (158). These data differ from the ones of Soetens et al. (392), who observe only polyubiquitylation of Gap1 in the presence of NH4+ and could not demonstrate an essential role of polyubiquitylation in sorting into the vacuolar pathway. Further experiments will be required to clarify the role of polyubiquitylation and monoubiquitylation in the control of Gap1 activity.

Like Gap1, the uracil permease Fur4 can be sorted directly from the biosynthetic pathway into the vacuolar MVB pathway. The signal for such sorting seems to be uracil, which triggers subsequent ubiquitylation and targeting of this permease into the lumen of the vacuole, where it is degraded (40).

B) GGA AND GOLGI TO ENDOSOME TRANSPORT.  GGAs [Golgi-localizing, {gamma}-adaptin ear domain homology, ADP-ribosylation factor (ARF)-binding proteins] are a family of monomeric adaptor proteins involved in membrane trafficking from the TGN to endosomes and lysosomes (42). They are ubiquitously expressed, and they are able to bind to clathrin in an ARF-dependent manner. GGAs are composed of a VHS (Vps27, Hrs, Stam) domain, a GAT [GGA and Tom1 (target of myb1)] domain, and an unstructured hinge region that connects to the COOH-terminal GAE ({gamma}-adaptin ear) domain (42) (Fig. 3). Such proteins mediate the sorting of mannose-6-phosphate receptors between the TGN and the endosomes. Recently, these proteins have been shown to bind to ubiquitin via the GAT domain (336, 369, 379), which binds the Ile-44 surface patch on ubiquitin. The residues of this patch are also necessary for the binding to Tsg101, the mammalian homolog of Vps23 (see below) (336). Knockdown of GGA3 by RNA interference (RNAi) leads to the accumulation of the mannose-6-phosphate receptors and of epidermal growth factor (EGF) receptor within large endosomes, likely the class E compartment (336). As mentioned above, Gap1 is sorted directly from the TGN to the vacuole when cells are grown on a good nitrogen source (158, 392), and this depends on polyubiquitylation of Gap1 (158). This sorting of Gap1 (or of Ste6) requires GGA2, which has been shown to interact with ubiquitin via its GAT domain. In {Delta}gga2 cells, or in cells expressing gga2 lacking a GAT domain, Gap1 is targeted to the plasma membrane, even in the presence of a good nitrogen source (369).


2. Ubiquitin and targeting to the vacuole

Once plasma membrane proteins are internalized they are either recycled back to the plasma membrane or they are targeted to the endosomal system, which coordinates protein transport from both the endocytotic and the biosynthetic pathways (143). A critical step in plasma membrane protein downregulation occurs in the sorting endosome, where MVBs are generated (111, 143). MVBs were described over 50 years ago (207 and references therein). They consist of a limiting membrane and up to several hundreds of internal vesicles of 40–90 nm in diameter. These are endosomal transport intermediates, which later mature or fuse with the late endosome compartment and subsequently with the lysosome/vacuole. They are generated when the limiting membrane in the sorting endosome invaginates and buds off toward the lumen of the organelle. During this process a subset of membrane proteins of the limiting membrane is sorted into these invaginating vesicles. Fusion of the organelle with late endosome, and ultimately the lysosome/vacuole, results in the delivery of the internal vesicles, along with the associated cargo, to the lumen of the lysosome/vacuole allowing their degradation by hydrolytic enzymes (123). Proteins that remain at the limiting membrane of endosomes remain also at the limiting membrane of lysosomes. This process, referred to as the MVB sorting process, serves different functions. Transmembrane proteins in the intraluminal membranes are susceptible to lysosomal degradation, whereas proteins in the limiting membrane are not, as they are exposed to the proteolytic environment only with their small luminal side, which often is proteolytically resistant due to extensive glycosylation. Second, the MVB may represent a storage compartment for proteins that are destined for secretion into the extracellular space. Third, signaling from the limiting membrane is in principle still possible, where this is precluded from receptors in the intraluminal membranes. We are now just beginning to understand the molecular mechanisms underlying protein sorting into the MVB pathway. This involves ubiquitin modification and recognition of the different ubiquitin binding domains/motifs. Importantly, these mechanisms are highly conserved between yeast and mammalian cells.

A) SORTING TO THE MVB PATHWAY.  As indicated above, both cell biological studies in mammalian cells and genetic investigations in yeast have been used to characterize the MVB pathway. Emr and collaborators (26) have identified so-called Vps (vacuolar protein sorting) mutants, which are defective in vacuolar sorting (26). More than 50 corresponding proteins have been found to function at different stages of protein sorting to the vacuole; 17 of them are required for targeting to the MVB pathway, and loss of function of these proteins leads to the formation of a malformed, very large late endosome/MVB compartment (349).

Many proteins that are sorted into the MVB are monoubiquitylated, whereas others, such as the Gap1 permease, are polyubiquitylated (158). This includes both newly synthesized hydrolytic enzymes, which are targeted via the TGN to the endosomes and ultimately to the lysosome/vacuole, but also the plasma membrane proteins described above that are internalized via ubiquitin-dependent endocytosis. Continuous ubiquitylation is necessary and sufficient to target these proteins to the endosomal/lysosomal lumen, as demonstrated by EGF receptor, which is recycled if its ubiquitylation is not maintained (252) (see below), or by fusion of a ubiquitin polypeptide to a transmembrane protein (346). However, Reggiori et al. (346) also demonstrated that not all proteins require ubiquitylation to enter the MBV pathway.

B) RECEPTORS RECOGNIZING UBIQUITYLATED CARGOES.  There have been three ubiquitin receptors described on endosomal membranes that recognize ubiquitylated cargo, namely, the Vps27/HseI (35, 381); the ESCRT-I (Fig. 4), which is composed of the three proteins Vps23, Vps28, and Vps37 (21, 38, 206); and the ESCRT-II complex comprising the proteins Vps22, Vps25, and Vps36 (17). The three complexes appear to bind monoubiquitylated cargoes via known ubiquitin interacting motifs or domains, namely, the UIM motif in Vps27/HseI (35), the UBC-like UEV domain in ESCRT-I (206), and the NZF domain in Vps36 (5). These binding motifs or domains have relatively low affinity for ubiquitin (115, 128, 333, 340, 374, 381, 412) but often appear in multiple copies on a protein, which likely lead to increased avidity. Different ubiquitin binding proteins may also cooperate in binding to ubiquitylated proteins, as suggested for Vps27 and Vps23 (20, 36, 208), which appear to interact via a PSDP motif at the COOH terminus of Vsp27. This PSDP motif resembles the PT/SAP motif in the HIV Gag protein, which is involved in binding to Tsg101 (the Vps23 mammalian homolog), and thereby recruiting the ESCRT machinery (128) (see below).


Figure 4
View larger version (61K):
[in this window]
[in a new window]
  		FIG. 4. Endosomal sorting and the MVB pathway. A: a model showing yeast ubiquitin-binding proteins and class E proteins and their role in MVB sorting of ubiquitylated cargoes. GGA, HseI, and Vps27 function together with clathrin upstream of the ESCRT complexes to recognize and concentrate ubiquitylated cargoes. Vps27 is recruited to the endosomal membrane via phosphatidylinositol 3-phosphate interaction and promotes assembly of ESCRT-I at the endosomal membrane. This activates the ESCRT-II complex, which promotes oligomerization of ESCRT-III. ESCRT-III recruits Bro-1 and the AAA-ATPase Vps4. The function of Bro-1 is to bind Doa4, which deubiquitylates cargoes. Vps4 then dissociates the ESCRT complexes from the endosomal membrane. Cargoes are subsequently recruited into the invaginating membrane for budding as multiple vesicles. B: mammalian homologs of the Vps proteins (indicated in the same colors) and their role in retrovirus budding. Retroviral Gag proteins can become ubiquitylated by Nedd4/Nedd4-like proteins if they contain a PY motif. Alternatively, they interact directly with their PTAP motif with the UEV domain of Tsg101, or they interact with the YPXL motif with AIP1/Alix1. This eventually leads to viral budding, as detailed in the text. Human homolog of Vps27: Tsg101; Vsp36: EAP45; Vps25: EAP25; Vps22: EAP30; Vps20: CHMP6; Vps2: CHMP2; Vps24: CHMP3; Snf7: CHMP4; Bro1: AIP1/Alix.

 
C) PROTEIN COMPLEXES INVOLVED IN TARGETING TO THE MVB PATHWAY.  Genetic evidence suggests that Vps27/HseI acts upstream of ESCRT-I (207). Lipids appear to play an important role in binding Vps27 via its FYVE (Fabl, YOTB, Vac1, and early endosomal antigen 1) domain to phosphatidylinositol-3-phosphate (PtdIns-3P)-enriched endosomal membranes (132, 208, 399). The lipid that is generated by phosphorylation of PtdIns by a class III PtdIns-3P kinase is found both on the limiting and intraluminal membranes (131). Further support for a role of phospholipids comes from the demonstration that the Fab1 PtdIns-3P 5-kinase is essential for the MVB pathway (305). Once Vsp27, which most likely is constitutively associated with HseI, is attached to the endosomal membrane (35), it recruits Vsp23 and the other components of the ESCRT-I complex (36, 208). This likely enables the two complexes to interact with the ubiquitylated cargo via the UIM motif in Vps27 and HseI and the UEV domain on Vps23.

Two other complexes downstream of ESCRT-I, namely, ESCRT-II and ESCRT-III, participate in protein sorting (Fig. 4). ESCRT-II is a ~155-kDa soluble complex composed of three proteins Vps22, Vps25, and Vps36 (17). The crystal structure of the core structure of the ESCRT-II complex was solved (165, 420, 457) and shown to comprise a quaternary complex of one Vps22, one Vps36, and two Vps25 molecules. A similar complex has also been described in mammalian cells (196). It forms a three-lobed or Y-shaped complex, with one Vps25 molecule that forms the stem and the other forming one arm. The second arm is composed of a sandwich comprising Vps22 and the COOH terminus of Vps36. Although the proteins have no homology to each other, they all form a pair of winged helices that together build the core of the complex. Such a structural block has been found previously in transcription factors. At the tip of the Vps36/Vps22 arm protrudes the NH2-terminal coiled-coil domain of Vps22. It was proposed that this NH2-terminal coiled-coil domain could interact with the coiled-coil region of Vps20 (in the ESCRT-III complex). However, Teo et al. (420) showed by binding assays that although Vps20 is involved in ESCRT-II/ESCRT-III interaction, the coiled-coil region of Vps22 is not necessary. Rather it is Vps25 that interacts with ESCRT-III Vps20 (420, 480). The coiled-coil domain may be involved in interaction with ESCRT-I, as suggested by complementary two-hybrid analysis (49), or play a role in membrane binding. An NH2-terminal ubiquitin binding NZF domain from Vps36 (not crystallized in either complex) is attached to this arm through a flexible linker (165). The domain may be involved in the transfer of ubiquitylated cargo from ESCRT-I to ESCRT-II. Indeed, mutations in the NZF domains of Vps36 lead to a class E compartment phenotype and to abrogation of CPS import into the endosomal lumen (5). However, the mammalian orthologs of Vps36 do not contain a NZF domain but interact via a novel ubiquitin binding domain, Glue, with ubiquitin (388). The protein complex transiently associates with endosomal membrane; this association may be promoted via direct interaction of the two Vps36 NZF domains with the ubiquitylated cargo, via interaction between ESCRT-I and ESCRT-II (49), or a combination of both.

ESCRT-II is important for the recruitment of ESCRT-III to the endosomal membrane; ESCRT-III is a large complex (Fig. 4) composed of four small proteins each containing a coiled-coil domain and an NH2-terminal basic region, whereas the COOH terminus is acidic (16). The four proteins are Vps20p, Snf7p, Vps2p, and Vs24p. In the cytosol they exist as soluble monomers. At the membrane they exist as two subcomplexes with different functions. Vps20 and Snf7 are required for the localization to the membrane and recruit Vps2/Vps24. The membrane localization of the complex is facilitated by myristoylation of Vps20. As outlined above, the complex is likely recruited by interaction between Vps25 (ESCRT-II) and Vps20. Snf7/Vps20 are required for membrane localization of Vps2/Vps24. These proteins apparently are the docking sites for Vps4, Bro1, and Doa4 (10).

D) VPS4.  Vps4 is a member of the AAA (ATPase associated with a variety of cellular activities) family of ATPases, which are involved in a number of different cellular functions such as membrane fusion (Sec18/NSF, cdc48/p97), protein degradation (Yta10–12, proteasome subunits), and chaperone-like activities (Yta10–12, Fsh) (for a review, see Ref. 68). The mammalian homolog of Vps4 is SKD1 (362). Vps4 has been shown to be essential for sorting of a number of proteins into the endosomal lumen (17–19, 113, 385, 483). Vps4 appears to be important for the release of the ESCRT complexes from the endosomal membrane, by yet unknown mechanisms (208). By analogy to Sec18/NSF, which disassembles coiled-coil complexes of SNARE proteins on membrane surfaces, it is possible that Vps4 also disassembles interactions between coiled-coil regions in the ESCRT complexes (16).

E) DOA4.  The deubiquitylating enzyme Doa4 is involved in the internalization/degradation of a number of proteins. As mentioned above, it is identical to Npi2, a protein essential for nitrogen catabolite inactivation of the Gap1 permease. Doa4 plays an essential role in ubiquitin homeostasis, but it has also a more specific role in endosomal targeting of the uracil permease Fur4. In pep4 cells, which lack vacuolar protease activity, nonubiquitylated Fur4 accumulates in the vacuole. In contrast, pep4 cells lacking Doa4 accumulate ubiquitylated Fur4, suggesting that Doa4 deubiquitylates Fur4 before entering into the vacuole (98). The data from Dupré et al. also suggest that numerous membrane proteins have to be deubiquitylated by Doa4 before the entry into the vacuole, as indicated by the accumulation of ubiquitylated membrane proteins in Doa4{Delta}Pep4{Delta} cells. This is further supported by findings that proteins involved in vacuolar import can act as suppressors of Doa4 cells (10). On the other hand, it has been shown that fusion of a ubiquitin-moiety on the transmembrane protein Phm5p bypasses the requirement of Doa4 activity for sorting into MVBs (346).

F) BRO1.  Bro1 (BCK1-like resistance to osmotic shock) was originally identified in a study showing that a bro1 mutation compromises the viability of cells that are mutant for several components of the protein kinase C/mitogen-activated protein (MAP) kinase signaling pathway, including the MEK kinase Bck1 (301). The BRO1 gene was also isolated as suppressor of mutations (ssy1 and ptr3) that decrease the capacity of a cell to uptake amino acids (117). Moreover, Bro1 is identical to Npi3, which is required for NH4+-induced downregulation and efficient ubiquitylation of Gap1 (398). Bro1/Npi3 is also the homolog of Vps31, involved in MVB sorting (305). In vps31 cells, both CPS and Ste2 are missorted to the vascular membrane or the class E compartment (258, 305, 306). Snf7 is necessary for recruitment of Bro1 to the class E compartment, but Vps24 is not required (306). Bro1 contains a coiled-coil (CC) region. This region is required for proper sorting of CPS into the lumen (258), but it is not necessary for targeting Bro1 to the endosomes. Bro1 interacts with Doa4, preferentially when associated with a membrane fraction. However, this interaction is not via the coiled-coil domain. Overexpression of Doa4 can suppress the sorting defect of either Bro1{Delta}CC or Bro{Delta}1 cells (with respect to the missorting of CPY), but can only partially suppress the sorting defect of CPS in Bro1{Delta} cells. Also, Doa4 suppresses the deubiquitylation defect of CPS in Bro1{Delta}CC cells. In addition, Doa4 localization to the endosomes requires Bro1, and the ESCRT-III protein Snf7, suggesting that Bro1 recruits Doa4 to the plasma membrane (258).

Bro1 has a mouse homolog, Aip1, which is localized both to the cytoplasm and cell membranes and interacts physically with Alg2, a calcium binding protein involved in cell apoptosis (281, 443). Interestingly, Aip1 also interacts with SETA/CIN85, an SH3-containing adapter shown to bind c-Cbl, which is involved in ubiquitylation and intracellular trafficking of receptor tyrosine kinase (see below). Moreover, it binds also to the CHMP4 proteins, the homologs of Snf7. Aip1 may thus be part of a protein complex involved in receptor ubiquitylation and/or trafficking, and it is involved in the MVB pathway.


   IV. MAMMALIAN MEMBRANE PROTEINS AND UBIQUITYLATION
Top
 Previous
 Next
 References
 
In recent years, a large number of studies have described the regulation of mammalian transmembrane proteins by the ubiquitin system. Key examples, representing various families of such proteins, are provided here.

A. Ion Channels

The regulation of ion channels by the ubiquitin system was described earlier for CFTR, where its processing in the ER was shown to be regulated by ubiquitylation and by ERAD (187, 453). Perhaps the best studied ion channel in which ubiquitylation plays a key role in regulating its cell surface stability is the epithelial Na+ channel (ENaC).


1. Regulation of ENaC by Nedd4 proteins

ENaC, comprised of three subunits ({alpha}beta{gamma}) (53, 54), is responsible for salt and fluid absorption in several epithelia including those in the kidney, colon, and lung. Each of the channel subunits contains a PY motif (PPxY) at its cytoplasmic COOH terminus (363), and deletion or mutation of the PY motif in beta- or {gamma}-ENaC causes Liddle's syndrome (149, 384), a hereditary hypertension caused by increased abundance and activity of ENaC at the plasma membrane (114). The same PY motifs were identified as binding partners for Nedd4 proteins (403). As depicted in Figure 2B, Nedd4 family members comprise a C2 domain responsible for membrane binding and subcellular localization (327, 449), three or four WW domains that bind PY motifs (L/PPxY) in target proteins (61, 200, 202), and a ubiquitin ligase HECT domain (173). Subsequent studies have demonstrated that Nedd4–2 (closely related to Nedd4–1) can suppress ENaC activity by controlling cell surface stability of the channel (Fig. 5) and that Liddle syndrome mutations impair the ability of Nedd4–2 to downregulate ENaC (3, 153, 197). In accord, cell surface stability of ENaC was shown to be regulated by ubiquitylation (404), and it was demonstrated that Nedd4–2 acts in concert with the E2