Distinct Functional Surface Regions on Ubiquitin

The characterized functions of the highly conserved polypeptide ubiquitin are to target proteins for proteasome degradation or endocytosis. The formation of a polyubiquitin chain of at least four units is required for efficient proteasome binding. By contrast, monoubiquitin serves as a signal for the endocytosis of plasma membrane proteins. We have defined surface residues that are important for ubiquitin's vital functions in Saccharomyces cerevisiae. Surprisingly, alanine scanning mutagenesis showed that only 16 of ubiquitin's 63 surface residues are essential for vegetative growth in yeast. Most of the essential residues localize to two hydrophobic clusters that participate in proteasome recognition and/or endocytosis. The others reside in or near the tail region, which is important for conjugation and deubiquitination. We also demonstrate that the essential residues comprise two distinct functional surfaces: residues surrounding Phe(4) are required for endocytosis, whereas residues surrounding Ile(44) are required for both endocytosis and proteasome degradation.

The characterized functions of the highly conserved polypeptide ubiquitin are to target proteins for proteasome degradation or endocytosis. The formation of a polyubiquitin chain of at least four units is required for efficient proteasome binding. By contrast, monoubiquitin serves as a signal for the endocytosis of plasma membrane proteins. We have defined surface residues that are important for ubiquitin's vital functions in Saccharomyces cerevisiae. Surprisingly, alanine scanning mutagenesis showed that only 16 of ubiquitin's 63 surface residues are essential for vegetative growth in yeast. Most of the essential residues localize to two hydrophobic clusters that participate in proteasome recognition and/or endocytosis. The others reside in or near the tail region, which is important for conjugation and deubiquitination. We also demonstrate that the essential residues comprise two distinct functional surfaces: residues surrounding Phe 4 are required for endocytosis, whereas residues surrounding Ile 44 are required for both endocytosis and proteasome degradation.
Conjugation to ubiquitin targets cellular proteins for degradation through two major pathways. Ubiquitination is required for the selective proteolysis of many intracellular proteins by the 26 S proteasome (1). In addition, a number of plasma membrane proteins require ubiquitination to signal their internalization or sorting in the endocytic pathway and subsequent degradation in the lysosome (2)(3)(4). Modification of substrates with ubiquitin proceeds through a three-enzyme cascade, resulting in the formation of a covalent isopeptide bond between the C-terminal glycine 76 of ubiquitin and a lysine residue in the target protein. Some substrates are conjugated to just one ubiquitin, whereas others are conjugated to multiple ubiquitins in the form of a polyubiquitin chain. Chains connected through three of ubiquitin's seven lysine residues (Lys 29 , Lys 48 , and Lys 63 ) have been identified in vivo in Saccharomyces cerevisiae (5)(6)(7)(8). Others have been synthesized in vitro and may exist in higher eukaryotic cells (9).
Efficient recognition of ubiquitinated substrates by the 26 S proteasome requires a minimum targeting signal consisting of four ubiquitin moieties linked to each other through isopeptide bonds between Gly 76 and Lys 48 (5,10). The Leu 8 , Ile 44 , and Val 70 amino acids in the ubiquitin polypeptide, known collectively as the hydrophobic patch, are critical for proteasomal degradation, although mutations in these residues have little effect on the formation of ubiquitin conjugates (11). Mutations of ubiquitin tail residues that disrupt ubiquitin conjugation can also affect proteolysis (12)(13)(14).
Ubiquitination regulates the endocytic traffic of signaling receptors, ion channels, permeases, and transporters in yeast and mammalian cells (reviewed in Refs. 2, 3, 15, and 16). In mammalian cells, the down-regulation of growth factor receptors and the epithelial sodium channel by endocytosis and degradation in the lysosome is regulated by ubiquitination. In the yeast Saccharomyces cerevisiae, G protein-coupled signaling receptors (Ste2p and Ste3p) as well as permeases and transporters are internalized into the endocytic pathway by a ubiquitin signal. Ligand binding stimulates phosphorylation and ubiquitination of Ste2p and Ste3p cytoplasmic tails, and both modifications are required for rapid receptor internalization (17)(18)(19). Receptors that lack post-translational ubiquitination sites in their cytoplasmic tails can be internalized by monoubiquitin fused in-frame (20,21). Monoubiquitin also promotes internalization of plasma membrane proteins in mammalian cells when it is fused to an N-terminal cytoplasmic domain (22). The single fused ubiquitin carries within its threedimensional structure all of the information necessary for regulated endocytosis (21). In contrast, monoubiquitin is an inefficient signal for proteasome recognition (10).
Ubiquitin is important for other cellular functions. Monoubiquitination of histones plays a role in meiosis in yeast and the development of Drosophila embryos (23,24), and monoubiquitination of a Fanconi anemia protein is linked to DNA repair and localization to nuclear foci (25). Monoubiquitination of the retroviral Gag protein is required for a late step in virus budding (26 -28). Independent of the proteasome, the formation of polyubiquitin chains linked through Lys 63 on the L28 ribosomal protein regulates ribosome activity (29), and Lys 63 -linked ubiquitin chains activate the IB kinase (30). These functions of ubiquitin have recently been discovered, and little is known about how they regulate protein structure and/or activity.
Although ubiquitin is a small protein, it is complex compared with other post-translational modifications such as phosphorylation and acetylation. In this study, we have begun to define the important functional features on ubiquitin's surface. First, we have identified the ubiquitin amino acids that are essential for vegetative growth of a yeast cell. These residues map to the ubiquitin tail and to two distinct functional faces of the ubiquitin globular domain. Second, we demonstrate that distinct amino acids of ubiquitin mediate its two best characterized functions. Whereas Ile 44 and surrounding hydrophobic residues are required for both proteasome recognition and endocy-tosis (11,21), a distinct surface region of ubiquitin containing Phe 4 is required only for endocytosis.
The plasmids containing STE2 and STE2-UBI variants were transformed into ste2⌬ strains by single-step gene transplacement at the ura3 locus. All mutant Ste2p and Ste2p-Ub 1 proteins were able to restore mating and ␣-factor binding in the ste2⌬ parental strain. The expression of Ste2p-Ub mutant proteins was assessed by immunoblot-ting using anti-Ste2p polyclonal antibodies. Two individual transformants of each mutant were assayed for their ability to internalize ␣-factor, and in each case, both transformants demonstrated similar internalization kinetics.
␣-Factor Internalization Assays-All ␣-factor internalization assays with strains expressing Ste2p-Ub3xR chimeras were performed as described (32,35). Internalization half-times were calculated based on exponential curve fits performed with Kaleidagraph Software (Synergy Software, Reading, PA). Each curve was fitted to a time course assayed for 60 min (seven data points), a best fit to the data was confirmed by visual inspection, and the half-times of internalization were determined from the average rate constants for two or three independent assays. Immunoblotting with anti-Ste2p antiserum confirmed that the mutations that caused a defect in endocytosis did not lead to proteolytic clipping of the receptor-ubiquitin chimera (data not shown).
Ubiquitin Conjugation and Proteasome Assays-Ubiquitination and degradation of 125 I-lactalbumin was assayed in rabbit reticulocyte fraction II (11). Lys 48 -linked Ub 4 (36) was used as an inhibitor or as a substrate for conjugation to a slightly modified form of H 10 -UbDHFR (10); specifically, the C-terminal HA epitope was replaced with a consensus site for protein kinase A phosphorylation (this site was not used in the present study). Wild type, F4Y, and L8A,I44A ubiquitin were expressed and purified as previously described (11) and assembled into Lys 48 -linked chains using E2-25K (37) under conditions optimized to provide similar length distributions. Prolonged incubation of E2-25K FIG. 1. A, ubiquitin amino acid sequence and comparison with the ubiquitin-like proteins Rub1p and Smt3p. Ubiquitin surface residues are depicted in black, and buried residues are shown in light gray. Red amino acids are the ubiquitin surface residues essential for vegetative growth of yeast. Residues conserved between the three polypeptides are highlighted in gray boxes. B, ubiquitin residues required for life in yeast. The three-dimensional structure of ubiquitin (42,43) is presented in images constructed with RasMol molecular visualization software (54). Essential ubiquitin amino acids are shown in blue. All essential residues are visible; none lie on the hidden surface. C, internalization information carried by the ubiquitin molecule. Primary residues required for internalization, Phe 4 and Ile 44 , are shown in magenta; residues that play a minor role in endocytosis are shown in pink.
with ubiquitin leads to the accumulation of circular polyubiquitin chains that are inefficiently recognized (38). Circular chains were absent from the preparations used here, based on the finding that Ͼ85% of the Ub 4 and Ub 5 in each mixture was disassembled to Ub 1 upon incubation with purified isopeptidase T. Proteasome assays employed 100 nM Ub 5 DHFR and 10 nM proteasomes (10), plus 0.5 mg/ml monoubiquitin to prevent absorptive loss of Ub 5 DHFR. Degradation was monitored by Western blotting with an antibody against the poly-His tag of UbDHFR (Santa Cruz Biotechnology Inc., Santa Cruz, CA); immunocomplexes were detected by colorimetric staining. Wild type and F4Y polyubiquitin chains were added to assays in an amount (6.8 mg/ml total ubiquitin) estimated to contribute ϳ13 M Ub 4 . The binding of mutant ubiquitins to purified E1 was monitored by a previously described thiol ester competition assay (39).
Analysis of the Ability of Mutant Ubiquitins to Support Growth of S. cerevisiae-Strains analyzed for sole source expression of mutant ubiquitin variants were derived from SUB328 (8) provided by Dan Finley (Harvard Medical School). This strain carried UBI on a LYS2-marked plasmid. SUB328 was transformed with UBI on a URA3-marked plasmid (LHP585) and mutant ubiquitin variants of this plasmid. These strains were propagated on medium containing ␣-aminoadipate at 24 or 30°C. ␣-Aminoadipate is converted to a toxic compound by the LYS2 gene product; therefore, only strains that were viable in the absence of a LYS2-UBI plasmid could survive (40). Strains viable on ␣-aminoadipate were further analyzed on selective SD minimal media to confirm loss of the LYS2-based plasmid and the presence of the URA3-based mutant ubi plasmid. Temperature sensitivity and cold sensitivity of viable strains carrying mutant ubiquitins as the sole source of ubiquitin were analyzed for single colony growth on YPUAD rich medium plates followed by incubation at the appropriate temperature.

RESULTS
Ubiquitin Residues Essential for Life in Yeast-Ubiquitin is highly conserved in eukaryotes, with only three amino acid differences between the yeast and human polypeptides at positions 19, 24, and 28. To identify surface residues (see Fig. 1A) required for ubiquitin's known and unknown essential functions, we performed a comprehensive alanine scan of the ubiquitin surface. Plasmids encoding the mutant ubiquitin proteins were introduced into a S. cerevisiae strain in which all endogenous ubiquitin genes were deleted and wild type ubiquitin was supplied on a plasmid (8). A plasmid shuffle was then used to assess the ability of each mutant ubiquitin to sustain life as the sole source of ubiquitin, as assayed by growth of strains at 24°C. Ubiquitin mutants that passed this test were further evaluated for temperature sensitivity at 37°C and cold sensitivity at 16°C. Strikingly, the majority of ubiquitin surface residues were not essential under our test conditions, and very few mutants exhibited conditional growth phenotypes ( Table I).
Sixteen of the 63 surface residues were essential for life in S. cerevisiae. When we mapped the essential ubiquitin amino acids onto the three-dimensional structure of ubiquitin, we found that the surface residues essential for viability clustered around two distinct patches in the globular domain and to the ubiquitin tail (Fig. 1B).
Internalization Information Carried by the Ubiquitin Polypeptide-The majority of ubiquitin's essential residues map to the C-terminal tail, which is involved in conjugation and deubiquitination, and to two surface patches near Phe 4 and Ile 44 . The Leu 8 /Ile 44 /Val 70 hydrophobic patch is required in the context of a polyubiquitin chain for proteasome binding and degradation (11). The key residues that are critical for the function of monoubiquitin to promote endocytosis have been identified as Ile 44 and Phe 4 (21). To refine our information about the functionally important domains of ubiquitin, we have thoroughly defined the role of ubiquitin surface amino acids in endocytosis. To do this, we analyzed scanning alanine mutants Wild type Ub

Functions of Ubiquitin Surface Domains in Yeast
of all of the surface residues of yeast ubiquitin made in the context of a Ste2p-ubiquitin chimera (35). Internalization of this chimeric protein is controlled by the fused ubiquitin and not by ubiquitin that is post-translationally conjugated to the chimeric protein, because the chimeric protein (Ste2p-Ub3xR) carries Lys 3 Arg substitutions at the known sites of polyubiquitin chain formation in vivo, Lys 29 , Lys 48 , and Lys 63 (7). Ubiquitin surface residues (see Fig. 1A) in Ste2p-Ub3xR were mutated individually or in clusters of 2-4 residues, and yeast strains expressing the mutant chimeras were assayed for their ability to internalize the Ste2p ligand, ␣-factor. Met 1 was not mutated; neither was Ser 28 , since this residue is an alanine in human ubiquitin. All of the receptor-mutant ubiquitin chimeras specifically bound a similar amount of ␣-factor at the cell surface, indicating that the chimeric proteins were transported to the plasma membrane. ␣-factor internalization by each mutant was measured in two or three independent assays, and curves were fitted to the data with computer software (see "Experimental Procedures"). The half-time of internalization for each mutant chimera was determined from the rate constants for the fitted curves (Table II). To assess the dynamic range of the internalization assay, we used a mutation (V26G) that disrupts the hydrophobic core of ubiquitin (41). This mutation inhibits internalization significantly (3.2-fold increase in half-time), demonstrating that ubiquitin's three-dimensional structure is important for its function as an internalization signal (21). Because Ste2p-UbV26G carries the full ubiquitin sequence in a strongly destabilized form, we used this mutant as a negative control in all subsequent analyses.
As we previously described, two point mutations in surface residues, F4A and I44A, had as severe an effect on internalization as V26G, and several other surface mutations inhibited internalization 2-fold or less (Ref. 21 and Table II). Although these mutations showed a continuum of modest effects, it is noteworthy that most of the residues with the largest defects in this class (1.6 -2-fold) cluster near either Ile 44 or Phe 4 in the three-dimensional structure of ubiquitin. These amino acids are Gln 2 , Lys 6 , Leu 8 , Thr 12 , Glu 64 , and Val 70 . Together with Ile 44 and Phe 4 , these residues define two discrete surface regions in the ubiquitin globular domain that are important for endocytosis (Fig. 1C).
In addition to its globular domain, ubiquitin carries a tail of four residues, Leu 73 , Arg 74 , Gly 75 , and Gly 76 , through which ubiquitin is post-translationally conjugated to substrate proteins. Gly 75 and Gly 76 are essential for conjugation, but their deletion had no effect on internalization ( Fig. 2A, Table II). Therefore, the C-terminal glycines do not carry information needed for internalization. (This observation also indicates that internalization of the receptor-ubiquitin chimeras does not require the conjugation of the C terminus of the fused ubiquitin to a lysine residue of another protein, confirming that internalization of the mutant chimeras is mediated by the fused ubiquitin moiety alone.) When Leu 73 and Arg 74 were mutated individually, each mutation caused 2.3-fold inhibition (Table II). However, the deletion of the entire four-residue tail had a more modest, but still significant, effect of 1.7-fold ( Fig. 2A, Table II). At present, it is unclear why deleting the tail has a smaller effect than mutation of Leu 73 or Arg 74 to alanine. These data suggest that the ubiquitin tail plays a minor role in postconjugation steps of ubiquitin-mediated endocytosis. The ubiquitin residues in the two surface regions and the tail that function in endocytosis are highlighted in Fig. 1C.
Internalization is the first role to be discovered for Phe 4 . To further characterize the role of this amino acid in internalization we made a more conservative mutation, Phe 4 3 Tyr, and analyzed the ability of this mutant to direct internalization of ␣-factor by the corresponding chimera. The F4Y mutation had as severe an effect on internalization as F4A (Fig. 2B, Table II). Because changing Phe 4 to Tyr adds a single hydroxyl group at a surface-exposed position, it is unlikely that this mutation inhibits internalization by disrupting the ubiquitin structure (this possibility was rigorously excluded; see below). Instead, these results suggest that Phe 4 may be involved in specific protein-protein interactions that facilitate endocytosis. Phe 4 Is Not Required for Ubiquitin Conjugation or Proteasomal Degradation-To assess the role of the Phe 4 patch and the Leu 73 and Arg 74 tail residues in conjugate formation and proteasome degradation, Q2A, F4A, F4Y, K6A, T12A, L73A, and R74A mutant ubiquitins were expressed in Escherichia coli, purified to homogeneity, and tested in ubiquitin-depleted reticulocyte lysates for their ability to support ubiquitination and degradation of 125 I-lactalbumin (11). Q2A, F4Y, F4A, and T12A ubiquitin behaved equivalently to wild type with respect to the rate of degradation (Fig. 3A, solid bars) and showed similar (or increased) levels of ubiquitin-lactalbumin conjugates in the steady-state (hatched bars). These results indicate that the Phe 4 patch of ubiquitin is not required for proteasome recognition. K6A, L73A, and R74A ubiquitin showed a somewhat reduced competence for degradation (Fig. 3A). Thus, the Lys 6 , Leu 73 , and Arg 74 residues participate in proteasome degradation but are not critically important, similar to their role in endocytosis. The K6A protein showed a strongly augmented level of conjugates in conjunction with a reduced degradation rate, suggesting that polyubiquitin chains assembled from this mutant are poorly recognized (11). Conjugates of high molecular weight, which are favored substrates for the proteasome (10), were underrepresented in the K6A ubiquitin assays (data not shown), suggesting that the apparent proteasome recognition defect may be an indirect consequence of a reduced efficiency of long polyubiquitin chain assembly. The competence of all the mutant ubiquitins in conjugation (Fig. 3A) indicates that they are properly folded, and consistent with this conclusion, the F4A and F4Y mutants were found to bind to the E1 ubiquitin-activating enzyme with affinities indistinguishable from wild type ubiquitin (data not shown).
To demonstrate specifically that the Phe 4 residue is not involved in the binding of polyubiquitin to proteasomes, we used an assay that measured the ability of different mutant ubiquitin chains to inhibit proteasome degradation. Unanchored chains of four wild type ubiquitin moieties (Ub 4 ) inhibit the degradation of a model proteasome substrate, ubiquitinated dihydrofolate reductase (Ub 5 DHFR; Fig. 3B) (10). By contrast, ubiquitin chains carrying mutations that affect proteasome binding, L8A and I44A, do not efficiently inhibit Ub 5 DHFR degradation. To test how UbF4Y chains behave, wild type, F4Y, and L8A,I44A ubiquitins were individually assembled into unanchored polyubiquitin chains (Fig. 3C) and the three chain mixtures were compared as inhibitors of Ub 5 DHFR proteolysis. Chains assembled from wild type or F4Y ubiquitin inhibited degradation almost completely, whereas, as expected (11), L8A,I44A chains inhibited degradation more weakly (Fig. 3D). These data demonstrate that Phe 4 is not required for targeting to proteasomes mediated by polyubiquitin chains. The data also show that Phe 4 is not required for ubiquitin conjugation mediated by the enzymes used in these experiments.

DISCUSSION
The ubiquitin molecule comprises a compact globular domain that consists of a five-stranded ␤-sheet and an ␣-helix and a flexible tail formed by four protruding residues (42,43). Two surface regions that are important for the defined functions of ubiquitin in proteasome recognition and endocytosis have been mapped onto this structure. A hydrophobic patch including Leu 8 , Ile 44 , and Val 70 is required for proteasome degradation (11) and plays a critical role in endocytosis (21). Phe 4 and adjacent residues are important for the endocytic role of ubiquitin but do not function in proteasome binding and degradation. Although the C-terminal tail of ubiquitin is required for all known ubiquitin functions, the essential nature of this structural feature is due to its role in conjugation. The tail is not known to be important for downstream recognition in any ubiquitin signaling pathway.
The ubiquitin residues essential for vegetative growth of yeast lie on these same three surface features, namely the tail and two distinct clusters on the globular core domain centered around Phe 4 or the Leu 8 /Ile 44 /Val 70 hydrophobic patch (Fig.  1B). Many of the essential amino acids we identified are known to be involved in ubiquitination, deubiquitination, and/or proteasome-mediated degradation (see below), all essential cellular processes. Except for the C terminus, the essential surface ubiquitin residues do not correlate closely with surface residues conserved in both of the ubiquitin-like proteins Rub1p and Smt3p (Fig. 1A). Although the Leu 8 /Ile 44 /Val 70 hydrophobic patch is conserved in Rub1p, Rub1p does not efficiently signal endocytosis when fused to Ste2p. 2 Our results suggest that the 2 J. C. Jemc and L. Hicke, unpublished data. inability of Rub1p to signal endocytosis may be due to the absence of residues comprising the Phe 4 patch. Nedd8, the mammalian Rub1p homologue, can replace one of the ubiquitins in tetraubiquitin. This observation suggests that the inability of Rub1p/Nedd8 to signal proteasomal proteolysis may reflect the absence of enzymatic mechanisms for assembly into substrate-linked chains, rather than an inability of proteasomes to recognize Rub1p/Nedd8 chains (39).
The ubiquitin tail consists of the essential residues Leu 73 , Arg 74 , Gly 75 , and Gly 76 . Gly 75 and Gly 76 are important for ubiquitin conjugation and deubiquitination (44 -46). Arg 74 is essential even though it is not important for E1 interaction or ubiquitin conjugation. These residues may be important for deubiquitination (47) and possibly for proteasome recognition as well (13). Arg 74 and Leu 73 play a minor role in endocytosis.
The essential surface cluster including Leu 8 , Ile 44 , and Val 70 consists of nine amino acids that extend from the base of the ubiquitin tail up to Gly 47 and Lys 48 (Fig. 1B). All of these residues are probably involved in ubiquitin conjugation and/or proteasome degradation, and they may also be important for deubiquitination (48). Lys 48 is the major site of polyubiquitin chain formation that is necessary for proteasome degradation (5). The role of Gly 47 has not been characterized, but since it bridges Lys 48 and the Ile 44 hydrophobic patch, it may play a role in proteasome degradation. Arg 42 and Arg 72 are required for efficient activation of ubiquitin by the E1 ubiquitin-activating enzyme (14,39), and Ile 36 plays a role in conjugation and/or proteasome recognition (11). The function of Leu 71 has not been investigated, but since it lies adjacent to Ile 36 , Val 70 , and Arg 72 it is likely to play a role in conjugation or proteasome degradation.
The second essential cluster on the globular domain surface consists of residues Gln 2 , Phe 4 , and Thr 12 . Phe 4 is critical for endocytosis, and Gln 2 and Thr 12 play a minor role. The effect of the F4A and F4Y mutations is not due to structural instability, because these mutant variants are conjugated to substrates efficiently ( Fig. 2A) and are expressed as well as wild type ubiquitin in yeast cells (data not shown). Moreover, the F4A and F4Y mutant proteins are recognized by proteasomes in the context of polyubiquitin chains (Fig. 3). Although most of the Phe 4 patch residues involved in endocytosis are essential, endocytosis itself is not absolutely required for vegetative growth of yeast at 24°C (49,50). This suggests that the Phe 4 patch may be involved in other, non-proteasome-dependent functions of ubiquitin. We conclude that ubiquitin carries a distinct essential surface region that is important for endocytosis, and perhaps for uncharacterized functions of ubiquitin, but not for proteasome degradation.
Our observations demonstrate that the Ile 44 hydrophobic patch is critical for multiple functions of ubiquitin and suggest that Ile 44  binding requires a tetraubiquitin chain, whereas endocytosis is triggered by monoubiquitination. Therefore, the interaction of a specific binding partner with Ile 44 may depend on the type of ubiquitin modification. Although monoubiquitinated receptors are efficiently recognized for endocytosis, the actual endocytic signal may be more complicated than a single ubiquitin. For example, monoubiquitinated plasma membrane proteins such as Ste2p form multimers (51,52), potentially allowing the display of a unique multimeric binding surface composed of noncovalently linked, but closely associated, ubiquitin monomers. Alternatively, the location of the ubiquitinated proteins at the plasma membrane may influence binding partner selection.
Monoubiquitin functions as a signal for internalization not only in yeast but also in higher eukaryotes. A fused ubiquitin moiety lacking lysine residues is sufficient for the endocytosis of chimeric plasma membrane proteins expressed in mammalian cells (22). As in yeast, Ile 44 is important for ubiquitin-dependent internalization in animal cells (22). The function of Phe 4 in animal cell endocytosis has not been reported. Ile 44 is part of a ubiquitin sequence, DQQRL 43 I 44 , that is similar to the DKQTLL dileucine endocytosis signal that was identified in the CD3 subunit of the T cell receptor (53). Like Ile 44 , the mutation of the buried Leu 43 to Ala severely inhibits endocytosis in yeast and animal cells (21,22). DQQRLI alone can serve as a transferable linear internalization signal when appended to a heterologous protein in mammalian cells (22). However, DQQRLI as a linear peptide signal cannot promote endocytosis in yeast (21). Furthermore, Leu 43 is buried in the hydrophobic core of the folded ubiquitin polypeptide and is therefore unlikely to be accessible for binding to endocytic machinery as part of a linear dileucine internalization signal.
In conclusion, ubiquitin carries a limited number of essential surface residues with defined functions in proteasome-mediated degradation or endocytosis. All essential residues are in, or adjacent to, two distinct hydrophobic surface patches. One of these surfaces is multifunctional and is involved in proteasome degradation and receptor endocytosis. The second, smaller hydrophobic surface is required for endocytosis but not for conjugation or proteasomal degradation. Previous mutational analyses of ubiquitin targeted a limited number of residues, and, with a few exceptions, consequences were examined with only one or a few enzymes. The present study is the first comprehensive analysis of essential features of the ubiquitin surface. Despite the functional importance implied by ubiquitin's extraordinary conservation, the majority of ubiquitin surface residues can be mutated without affecting viability under laboratory conditions. Perhaps the nonessential residues provide an evolutionary advantage in the outside environment by participating in nonessential ubiquitin-dependent processes such as DNA repair, histone modification, or the regulation of ribosomal proteins (8,23,24,29).