Identification of Rabbit Reticulocyte E217K as a UBC7 Homologue and Functional Characterization of Its Core Domain Loop*

The structural basis by which ubiquitin (Ub)-conjugating enzymes (E2s) determine substrate specificity remains unclear. We cloned rabbit reticulocyte E217K because unlike the similarly sized class I E2s, E214K and UBC4, it is unable to support ubiquitin-protein ligase (E3)-dependent conjugation to endogenous proteins. RNA analysis revealed that this E2 was expressed in all tissues tested, with higher levels in the testis. Analysis of testis RNA from rats of different ages showed that E217K mRNA was induced from days 15 to 30. The predicted amino acid sequence indicates that E217K is a 19.5-kDa class I E2 but differs from other class I enzymes in possessing an insertion of 13 amino acids distal to the active site cysteine. E217K shows 74% amino acid identity with Saccharomyces cerevisiae UBC7, and therefore, we rename it mammalian UBC7. Yeast UBC7 crystal structure indicates that this insertion forms a loop out of the otherwise conserved folding structure. Sequence analysis of E2s had previously suggested that this loop is a hypervariable region and may play a role in substrate specificity. We created mutant UBC7 lacking the loop (ubc7Δloop) and a mutant E214k with an inserted loop (E214k+loop) and characterized their biochemical functions. Ubc7Δloop had higher affinity for the E1-Ub thiol ester than native UBC7 and permitted conjugation of Ub to selected proteins in the testis but did not permit the broad spectrum E3-dependent conjugation to endogenous reticulocyte proteins. Surprisingly, E214k+loop was unable to accept Ub from ubiquitin-activating enzyme (E1) but was able to accept NEDD8 from E1. E214k+loop was able to support conjugation of NEDD8 to endogenous reticulocyte proteins but with much lower efficiency than E214k. Thus, the loop can influence interactions of the E2 with charged E1 as well as with E3s or substrates, but the exact nature of these interactions depends on divergent sequences in the remaining conserved core domain.

The structural basis by which ubiquitin (Ub)-conjugating enzymes (E2s) determine substrate specificity remains unclear. We cloned rabbit reticulocyte E2 17K because unlike the similarly sized class I E2s, E2 14K and UBC4, it is unable to support ubiquitin-protein ligase (E3)-dependent conjugation to endogenous proteins. RNA analysis revealed that this E2 was expressed in all tissues tested, with higher levels in the testis. Analysis of testis RNA from rats of different ages showed that E2 17K mRNA was induced from days 15 to 30. The predicted amino acid sequence indicates that E2 17K is a 19.5-kDa class I E2 but differs from other class I enzymes in possessing an insertion of 13 amino acids distal to the active site cysteine. E2 17K shows 74% amino acid identity with Saccharomyces cerevisiae UBC7, and therefore, we rename it mammalian UBC7. Yeast UBC7 crystal structure indicates that this insertion forms a loop out of the otherwise conserved folding structure. Sequence analysis of E2s had previously suggested that this loop is a hypervariable region and may play a role in substrate specificity. We created mutant UBC7 lacking the loop (ubc7⌬loop) and a mutant E2 14k with an inserted loop (E2 14k ؉loop) and characterized their biochemical functions. Ubc7⌬loop had higher affinity for the E1-Ub thiol ester than native UBC7 and permitted conjugation of Ub to selected proteins in the testis but did not permit the broad spectrum E3-dependent conjugation to endogenous reticulocyte proteins. Surprisingly, E2 14k ؉loop was unable to accept Ub from ubiquitin-activating enzyme (E1) but was able to accept NEDD8 from E1. E2 14k ؉loop was able to support conjugation of NEDD8 to endogenous reticulocyte proteins but with much lower efficiency than E2 14k . Thus, the loop can influence interactions of the E2 with charged E1 as well as with E3s or substrates, but the exact nature of these interactions depends on divergent sequences in the remaining conserved core domain.
In eukaryotes, selective degradation of many key cellular proteins is mediated by the ubiquitin/proteasome system (re-viewed in Refs. 1 and 2). These proteins include gene transcription factors (3,4), mitotic cyclins (5-7), abnormal/mutated proteins (8), oncoproteins (9), the tumor suppressor protein p53 (10), and cell-surface receptors (11). Through degradation of these proteins, the ubiquitin/proteasome system is implicated in the regulation of diverse processes such as gene transcription, DNA repair (12), cell cycle, cellular stress response, receptor endocytosis (13), and antigen processing (14). How the ubiquitin/proteasome system recognizes a multitude of specific substrates remains to be elucidated and is a question of current interest.
Proteins recognized by this system are targeted to hydrolysis by the 26 S proteasome through the covalent attachment of ubiquitin, a highly conserved 76-amino acid protein. Thus, the ubiquitin-conjugating apparatus plays a key role in selection of substrates. The conjugation of ubiquitin to protein substrates is a multistep process involving three enzymes (15). The first step is catalyzed by ubiquitin-activating enzyme (E1). 1 In an ATP-hydrolyzing reaction, a ubiquitin adenylate intermediate is formed, followed by transfer of the C terminus of the ubiquitin to the thiol group of the active site cysteine residue of the E1 enzyme (16). This activated ubiquitin is then transferred from E1 to a specific cysteine residue of one of several ubiquitin-conjugating enzymes (E2s) (17). Some E2s transfer ubiquitin to ubiquitin-protein ligases (E3s) that bind substrates. This step also involves the formation of a thiol ester linkage between ubiquitin and the ligase (18). Other E2s appear to bind to E3s, and in these situations, the E3s appear to act as docking proteins, which bring together the E2 and the substrate (19). Some E2s can transfer ubiquitin to the substrate directly (20). The E2 and/or E3 enzymes finally catalyze isopeptide bond formation between the C terminus of ubiquitin and the ⑀-amino group of internal lysine residues of target proteins. In most cases, conjugation of ubiquitin to ubiquitin moieties already linked to the protein leads to formation of multi-ubiquitin chains attached to the substrate (2). These multi-ubiquitin chains are subsequently recognized by the 26 S proteasome. Proteins are then probably unfolded and translocated into the central cavity of the proteasome where they are degraded to small peptides (21).
A large body of genetic and biochemical evidence indicates that both E2 and E3 enzymes exist as protein families and together define the substrate specificity of the conjugation system (1,2). Ubiquitin-conjugating enzymes are a family of closely related proteins. Functionally, these have been best characterized in Saccharomyces cerevisiae, in which 13 E2 genes have been described. Inactivation of many yeast E2 genes produces specific phenotypes indicating that different E2s have different functions. For example, the near identical E2s, UBC4/UBC5 are involved in degradation of abnormal proteins, sporulation, and resistance to stress conditions (8). UBC6 is involved in degradation of the yeast transcriptional repressor MAT␣2 (22). UBC7 is responsible for resistance to cadmium toxicity (23), degradation of MAT␣2 (22), and certain proteins localized to the endoplasmic reticulum/nuclear envelope (24). This indicates that E2s are involved in determining substrate specificity either by interacting with specific E3s which in turn bind the substrates or less likely by directly recognizing and ligating ubiquitin to a target protein. E2 protein structures have been intensely studied because they may reveal some of the mechanisms underlying recognition of substrates. All E2s have a core domain of about 150 amino acids that shows at least 25% sequence identity. E2s can be divided broadly into four classes (17). Class I enzymes contain only the conserved catalytic core domain. Class II and III enzymes have either extra C-terminal or extra N-terminal extensions attached to the core domain, respectively. Class IV enzymes have both C-terminal and N-terminal extensions. These extensions to the core domain have been proposed to confer specificity for recognition of substrates or E3s (25) or to provide a localization signal (26). However, divergent sequences within the conserved core domain of E2 proteins may also be highly significant for recognizing substrates.
Haas and co-workers (27) initially identified five major E2s in reticulocyte extracts. E2 14K supports an apparently wide spectrum conjugation of ubiquitin to endogenous proteins (27) and was subsequently determined to be a class I E2 (28). Interestingly, the other prototype class I E2, the UBC4/5 subfamily, also supports this apparent wide spectrum conjugation of ubiquitin to endogenous proteins (29,30). E2 20K , E2 25K , and E2 32K were subsequently found to be class II E2s and to conjugate ubiquitin in vitro only to selected substrates (31). However, no apparent substrates have been identified for E2 17K , and its molecular structure remains unexplored. Although it is relatively similar in size to E2 14K (which actually has a molecular mass of 17.3 kDa) and therefore may also be a class I enzyme, it is clearly functionally distinct, having no identified substrates. Therefore, we decided to explore the molecular basis of this functional difference by characterizing this E2.
These primers were then used in RNA PCR reactions to amplify a cDNA fragment encoding part of E2 17K . The cDNA templates were synthesized from rat brain, bone marrow, liver, testis, and rabbit reticulocyte RNA (2 g) using reverse transcriptase (Life Technologies, Inc., Superscript Preamp Kit) according to the supplier's protocol and oligo(dT) as a primer. In the PCR reactions, annealing was carried out at 55°C, extension at 72°C for 1 min, and 30 cycles were performed.
The amplified DNA fragment was subcloned into pGEM-T (Promega) and sequenced using the dideoxy chain termination method and T7 polymerase (Amersham Pharmacia Biotech). Since the predicted protein sequence was identical with the sequence obtained from the two E2 17K peptides, the PCR-amplified DNA fragment was labeled with 32 P and used as a probe to screen a rat testis cDNA library in the zapII vector (Stratagene). An aliquot of the library containing one million recombinants was screened by transferring plaques to nylon membranes and hybridized with the probe. Positive phage clones were purified, and the pBluescript phagemid containing the insert was excised from the phage according to the manufacturer's instructions. The insert in the plasmid was sequenced. Sequences were analyzed using the Blast software at the National Center for Biotechnology Information.
RNA Analysis-Expression of rat UBC7 was evaluated either by RNA blotting (Northern analysis) or by RNase protection assays. RNA was prepared from various rat tissues by the guanidinium thiocyanate-CsCl method (33). RNA blotting was performed by resolving the indicated quantities of RNA on agarose gels containing 1% formaldehyde followed by transfer to nylon membranes and cross-linking with UV light. The membranes were hybridized with the indicated 32 P-labeled DNA probes, washed, and then subjected to autoradiography.
RNase protection assays for measuring rat UBC7 were performed as described previously (34). The RNA probe was the antisense strand of the region encoding amino acid residues 71-140, which was amplified by PCR and subcloned into pGEM-T. 32 P-Riboprobes encoding the complementary strand were synthesized in vitro by using the appropriate T7 or SP6 polymerase and were hybridized with RNA samples (10 g) overnight at 46°C. Subsequently, nonhybridizing RNA was digested with RNases A/T, and the remaining labeled RNA hybrids were resolved on a 4% acrylamide DNA sequencing gel. Size of the bands was estimated by running a sequencing reaction alongside the samples. To confirm that the band seen corresponded to that expected, RNA encoding the rat UBC7 mRNA was synthesized in vitro and similarly hybridized with the probe.
Recombinant Expression of UBC7, ubc7⌬loop, E2 17K , and E2 17K ϩ Loop-A UBC7 construct lacking the loop (PGEDKYGYEKPEE) was derived by a PCR-based mutagenesis method. To prevent misfolding of ubc7⌬loop without the loop, the residues adjacent to the loop were mutated to those in E2 14K . Two overlapping internal primers, one forward and one reverse, were synthesized and both contain the same deletion of nucleotides encoding amino acid residues 97-109 and same mutation of nucleotides encoding residue 95 from His to Gln; 96 from Glu to Asn (internal forward primer, 5Јcagaatagatggctgcctatccatac3Ј; internal reverse primer, 5Јtctattctgaagaatagaaatgcaaac3Ј). The internal reverse primer and the external forward primer (5Јacggagctg-cagtcggc3Ј) were used in one PCR reaction to amplify the 5Ј part of the E2. The internal forward primer and the external reverse primer containing a BamHI site (5Јggatccctcaattggagaccctg3Ј) were used in another PCR reaction to amplify the 3Ј part of the E2. The purified products of these reactions were then mixed together and amplified using only the external forward and reverse primers to produce the full-length product without the 39 nucleotides encoding the 13 extra amino acids. This product was digested with BamHI and subcloned into the pET-11d expression vector (Novagen) that had been digested with NcoI, filled in with Klenow polymerase, and then digested with BamHI. The E2 14K containing the inserted loop (E2 14K ϩloop) was amplified using the same principle as above except that the two overlapping primers contained the same insertion of nucleotides encoding the 13 amino acids (internal forward primer, 5Јaaatatggttatgagaagccagag-gaacgc3Ј; internal reverse primer, 5Јctcataaccatatttatcctctccaggttcag3Ј; external forward primer, 5Јtcgaccccggcgcggaggagg3Ј; external reverse primer, 5Јggatccgatggaaaggatcaacagc3Ј). The full-length product of E2 14K containing the loop was also subcloned into the pET-11d vector. The pET-11d plasmids containing UBC7, ubc7⌬loop, E2 14K , and E2 14K ϩloop were transformed individually into Escherichia coli BL21 (DE3). Following induction of expression of T7 polymerase by incubation of the bacteria with isopropyl-␤-thiogalactopyranoside for 2 h at 30°C, the cells were harvested, sonicated, and centrifuged at 10,000 ϫ g. The supernatants containing the individual E2s were used in the assays, or the E2s were purified as described below.
Immunoblotting of the E2 17K and E2 17K ϩLoop-Bacterial lysates containing E2 14K or E2 14K ϩloop were resolved on 15% SDS-PAGE gels and transferred to 0.1-m nitrocellulose membranes. Membranes were probed with anti-E2 14K antibody, followed by incubation with secondary antibody, 125 I-goat anti-rabbit IgG. The membranes were exposed to x-ray films and to a PhosphorImager screen for quantification. Phos-phorImager analysis was carried out using a Fuji imaging plate (Fuji Photo Film Co., Ltd.). Various amounts of the lysates were analyzed to confirm linearity of the quantitation.
Purification of Recombinant NEDD8 -A DNA fragment encoding the mature proteolytically processed form of rat NEDD8 was amplified from rat muscle RNA using reverse transcriptase-PCR, and oligonucleotides derived from the NEDD8 sequence (35) previously deposited in GenBank TM . The fragment was subcloned into the bacterial expression vector pET11-d (Novagen) and transformed into E. coli BL21 (DE3). Following induction of expression with 1 mM isopropyl-␤-thiogalactopyranoside for 2 h at 30°C, cells were harvested from 800 ml of culture and rinsed with phosphate-buffered saline, and the cell pellets were frozen at Ϫ20°C. Subsequent manipulations were on ice or at 4°C. The frozen pellets were resuspended in one-tenth of the original culture volume of 50 mM Tris, pH 7.5, 5 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, and 5 g/ml pepstatin A and lysed by sonication. The lysate was clarified by centrifugation at 100,000 ϫ g for 1 h, and then proteins were differentially precipitated by ammonium sulfate (ICN Ultrapure). The 60 -95% ammonium sulfate fraction containing the NEDD8 was dialyzed against 50 mM Tris, pH 7.5, and then passed over a 7-ml DEAE-cellulose column equilibrated in the same buffer. NEDD8 remained in the unbound fraction, and this fraction was then dialyzed overnight against 50 mM ammonium acetate, pH 5.0, before being applied to a 5 ϫ 100-mm propylsulfonic acid cation exchange column (Waters SP/HR15) equilibrated in the same buffer. Bound proteins were eluted with a 0 -0.5 M NaCl gradient (10 mM/min) in the same buffer. NEDD8 eluted at approximately 0.125 M NaCl and in most fractions was Ͼ99% pure as evaluated by Coomassie Bluestained acrylamide gels. Molecular mass determination by Maldi-Toff mass spectrometry confirmed that the purified NEDD8 peptide was intact.
Iodination of Proteins-The chloramine-T method was used to label bovine ubiquitin with Na 125 I to a specific radioactivity of 6,000 cpm/ pmol and to label NEDD8 to a specific radioactivity of 15,000 cpm/pmol. Unincorporated 125 I was removed by passing the reaction products over a Sephadex G-25 column.
Thiol Ester Assay-The relative enzymatic activities of E2 14K, E2 14k ϩloop, UBC7, and ubc7⌬loop containing bacterial lysates were quantitated by their abilities to form thiol ester complexes with 125 Iubiquitin. E1 used in the assays was prepared from rabbit reticulocytes or rabbit liver as described previously (27). Various amounts of E2s were incubated in a total volume of 10 l as follows: 50 mM Tris, pH 7.5, 1 mM dithiothreitol, 2 mM MgCl 2 , 2 mM ATP, 50 nM (unless otherwise indicated) E1, inorganic pyrophosphatase (20 units/ml), 5 M 125 I-ubiquitin (Ͼ3000 cpm/pmol) at 37°C for 1 min. The reaction was stopped with Laemmli sample buffer with or without 2-mercaptoethanol (to hydrolyze or preserve the thiol ester linkages, respectively) and resolved by SDS-PAGE on 12.5% acrylamide gels at 4°C. Following detection by autoradiography, the thiol ester bands were excised from the dried gel in order to measure the incorporated radioactivity and thereby estimate the E2 enzymatic activity. Assaying dilutions of the E2-containing extracts confirmed linearity of assays. The relative enzymatic activity of bacterial lysates containing E2 14K and E2 14k ϩloop to form thiol ester complexes with 125 I-NEDD8 was determined using the same method except that the concentration of 125 I-NEDD8 was 100 M, and the reaction time was 2 min to ensure full charging of the E2s with NEDD8.
To test the abilities of UBC7 and ubc7⌬loop to interact with E1, their apparent K m values for E1 charged with ubiquitin were determined. Recombinant UBC7 and ubc7⌬loop were purified by adding E1 to the bacterial lysates followed by ubiquitin affinity chromatography and quaternary amine anion exchange chromatography (Amersham Pharmacia Biotech MonoQ) as described previously (28). Varying concentrations (50 -500 nM) of the purified E2s were assayed in thiol ester reactions as above except that a limiting concentration of E1 was used (2 pM), and the time course of formation of the E2 thiol ester was monitored by removing aliquots from the reaction every 30 s over a 2-min period. Initial velocities were calculated from the time courses and used in double-reciprocal plots (Lineweaver-Burk) to determine apparent K m values and maximal velocities.
Conjugation Assays-The abilities of native and mutated forms of the E2s to support conjugation of ubiquitin or NEDD8 to endogenous proteins were tested as follows. The test fraction derived from endogenous proteins was incubated in a final volume of 25 l containing 50 mM Tris, pH 7.5, 1 mM dithiothreitol, 2 mM MgCl 2 , 2 mM ATP, 0.5 units of inorganic pyrophosphatase, 12.5 mM phosphocreatine, 2.5 units of creatine kinase, 50 nM E1 (100 nM when used with NEDD8), 5 M 125 Iubiquitin or 100 M 125 I-NEDD8 and the indicated concentrations of E2s. Following incubation for the indicated times at 37°C, the reactions were stopped with Laemmli sample buffer containing 2-mercaptoethanol, the products resolved of free ubiquitin or NEDD8 by SDS-PAGE on 10% acrylamide gels, and detected by autoradiography. When required, lanes were cut out and the incorporated ubiquitin or NEDD8 quantitated by counting in a gamma counter. The rabbit reticulocyte fraction devoid of ubiquitin-conjugating enzymes was prepared by a 30% ammonium sulfate precipitation of fraction II as previous described (27). Preparation of crude testis cytosol and its fractionation by quaternary amine anion exchange chromatography (Amersham Pharmacia Biotech MonoQ) was as described previously (34).

RESULTS AND DISCUSSION
Cloning of E2 17K -To determine whether E2 17K is a class I enzyme, we obtained peptide sequence from the purified protein for the purpose of molecular cloning. Direct sequencing of purified rabbit reticulocyte E2 17K protein did not yield any sequence suggesting the presence of a blocked N terminus. Therefore, the protein was digested with cyanogen bromide. Three peptides, two of approximately 6 kDa and one of approximately 5 kDa, were generated. Two peptide sequences were obtained, one of 17 residues and another of 16 residues (Fig. 1). By examining data bases, we found that these two peptides revealed significant sequence similarities with E2s in various species. To obtain the cDNA sequence, degenerate primers were designed encoding a part of each peptide sequence (Fig.  1). These primers were used with rabbit reticulocyte, rat testis, kidney, and muscle cDNAs in PCR reactions. A fragment of 207 bp was amplified from all of the tissue cDNAs. Sequencing of the fragments and translation of the sequences yielded identical amino acid sequences that corresponded to the peptide microsequencing results, confirming that the DNA fragment encodes part of E2 17K . Because RNA analysis with this cDNA FIG. 1. Alignment of rat UBC7 with other class I ubiquitin-conjugating enzymes. Rat UBC7 protein sequence was deduced from the cDNA clones and compared with S. cerevisiae UBC7 (YUBC7), UBC4 (UBC4), and to mammalian E2 14K . Asterisk indicates perfectly conserved residues; dot indicates well conserved residues; arrowhead indicates active site cysteine; bold residues indicate sequences of peptides obtained following cyanogen bromide fragmentation of purified rabbit E2 17K . Degenerate oligonucleotides encoding the underlined sequences were used to amplify the intervening sequence by reverse transcriptase-PCR. as probe (see below) showed higher expression of E2 17K in rat testis (Fig. 2), a cDNA library derived from this tissue was screened to obtain full-length clones. Five positive clones were found, all possessing an open reading frame encoding a protein of 170 amino acids and a predicted mass of 19.5 kDa. Comparison of the sequence with the E2s in the GenBank TM data base showed that this protein has 74% amino acid identity with S. cerevisiae UBC7 (Fig. 1). It has the conserved core catalytic domain but lacks N-or C-terminal extensions and is therefore a class I enzyme. The core domain contains the active site cysteine and the conserved "HPN" tripeptide found in all E2s. The deduced amino acid sequence was also found to be identical to human E2g (36), confirming again the conservation of the ubiquitin system. Since most investigators in this area are naming mammalian E2s after their apparent yeast homologues, we therefore now refer to E2 17K as mammalian UBC7.
Expression of UBC7 in Various Rat Tissues-To determine which tissues express UBC7, RNA from various rat tissues were analyzed by Northern blotting using the fragment of rat UBC7 cDNA as a probe. One transcript (2.5 kilobase pairs) was observed in all rat tissues examined. Thus, this E2 appears to have a general cellular function. This would be consistent with results from studies in yeast which indicate that UBC7 plays a role in the degradation of abnormal proteins in the endoplasmic reticulum (24). When normalized to 18 S rRNA expression, all tissues showed low expression levels, but slightly higher expression was seen in testis ( Fig. 2A). This is in contrast to the expression pattern of human E2g which when normalized for actin expression showed similar levels in all tissues (36). This is likely due to our use of tissues from young peripubertal rats for this analysis, whereas the samples of human mRNA in the E2g study were derived from an adult which in the rat has lower testis expression (see below). The more sensitive RNase protection assay confirmed that rat UBC7 mRNA is expressed diffusely with a slightly higher expression in testis (Fig. 2B).
Since the testis undergoes major postnatal development, we tested whether rat UBC7 expression was developmentally regulated. RNAs from testes of rats of different ages were analyzed by Northern blotting for the expression of UBC7 with the 207-bp fragment as a probe. UBC7 expression was low in testes from 10-day-old rats, increased in testes from 15-to 30-day-old rats, and then decreased upon further maturation to the adult testis (Fig. 3). The peak expression during days 15-30 of life suggests that UBC7 mRNA is present in spermatocytes and possibly early spermatids which make their first appearance in the testis during this period. The subsequent fall in expression in older animals suggests that the mRNA is less prominent or absent in more mature spermatids which begin to accumulate in the testis after 30 days of age. We have previously observed regulation of expression of other E2s during development of the testis, particularly the UBC4 isoforms. UBC4-testis is induced in spermatids (34); UBC4-1 is present in all cells but with higher expression in spermatocytes, and UBC4-2 shows maximal expression in elongated spermatids. 2 Thus, these E2s are likely induced at different stages to target-specific populations of proteins for degradation during this complex developmental process.
Comparison of UBC7 with Other E2s-To begin determining why UBC7, a class I enzyme, is functionally distinct from other FIG. 2. Expression of UBC7 mRNA in rat tissues. A, RNA samples (20 g) from the indicated tissues from 35-day-old rats were resolved by electrophoresis on 1% agarose gels and transferred to a nylon membrane. Following hybridization with a 32 P-labeled fragment from the UBC7 cDNA, the membrane was washed and exposed to film (top). Following removal of the probe, the membrane was rehybridized with a probe encoding the 18 S ribosomal RNA (bottom) to evaluate loading and transfer of the samples to the membrane. EDL, extensor digitorum longus; SOL, soleus; Testis(half), only 10 g of RNA loaded. B, a fragment (207 bp) of the UBC7 cDNA was subcloned into pGEM-T vector and used to synthesize an antisense 32 P-labeled RNA probe. RNA samples were prepared from the indicated rat tissues. Following hybridization of 10-g aliquots of the RNA samples with the 32 P-labeled riboprobes and digestion by a RNase A/T mixture, the protected fragments were resolved on a 4% denaturing polyacrylamide gel and detected by autoradiography. FIG. 3. Regulation of UBC7 mRNA expression during postnatal development of the testis. RNA (10 g) from rat testes of different ages were electrophoresed on a 1% agarose gel and transferred to nylon membrane. After hybridization with the 32 P-labeled UBC7 cDNA probe, the blot was subjected to autoradiography and quantitated by Phospho-rImager analysis. After stripping the previous probe, the blot was rehybridized with a probe encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH). A, representative lanes of blot. B, quantitation of all samples with mRNA expressed per unit of glyceraldehyde-3-phosphate dehydrogenase mRNA. There were significant differences between the means (p Ͻ 0.01) as determined by one-way analysis of variance.
class I E2s, such as E2 14K and mammalian UBC4/UBC5, we compared the structure of these E2s. Rat UBC7 has 30% amino acid identity with rabbit E2 14K . The alignment of the rat UBC7, S. cerevisiae UBC7, UBC4, and rabbit E2 14K required an insertion of 13 amino acids (Pro 97 -Asp 109 ) in rat UBC7 and S. cerevisiae UBC7. This region is close to the ubiquitin-accepting cysteine (UBC7 Cys 90 ) (Fig. 1). Since the crystal structures of S. cerevisiae UBC7 (37), UBC4 (38), and Arabidopsis thaliana UBC1 (39), an E2 14K homologue, have been determined, we also compared their three-dimensional structures. Their crystal structures are very similar. They all possess an antiparallel ␤-sheet with four strands bound on each end and on one side by four ␣-helices. The ubiquitin-accepting cysteine residue is located in a long extended stretch between the fourth strand of the ␤-sheet and the second ␣-helix. The major difference between S. cerevisiae UBC7 and the other E2s is the insertion containing the 13 extra residues. This insertion forms a large loop near the catalytic site, which alters one surface region but does not change the overall folding pattern of the conserved core domain. Comparison of this region in the sequences of many E2s suggested that it may represent a hypervariable surface in a common E2 tertiary fold. Thus this region was hypothesized to play an important role in specifying functions of the E2s (37). Since UBC7 is distinct from the other prototype class I E2s, such as E2 14K and UBC4, in being unable to support E3-dependent conjugation, we evaluated whether this is caused by this 13 extra residue loop in UBC7.
Characterization of Rat UBC7 and ubc7⌬loop-To address this question, the 13 extra residues were deleted from UBC7 to form ubc7⌬loop. Native rat UBC7 and ubc7⌬loop were expressed in E. coli. Thiol ester assays were performed to test if native UBC7 and ubc7⌬loop can accept ubiquitin from E1. In the presence of 125 I-labeled ubiquitin and purified E1, both UBC7 and ubc7⌬loop can form thiol esters with 125 I-ubiquitin (Fig. 4A). However, these end point assays are unable to detect more subtle differences in interaction with E1. Therefore we measured apparent K m values of UBC7 and ubc7⌬loop for the E1-ubiquitin thiol ester (Table I). Interestingly, removal of the loop decreased the K m value while maintaining similar V max values indicating that the loop can negatively influence binding of the E2 to the charged E1.
We next tested whether UBC7 and ubc7⌬loop can support E3-mediated ubiquitination to endogenous proteins of the reticulocyte extract. As was shown previously, E2 14K and UBC4 can efficiently support incorporation of ubiquitin into proteins, whereas native UBC7 does not (27). To confirm that this inability of UBC7 was not due to the higher apparent K m for charged E1 than E2 14k (31), we repeated the assay using a 10-fold higher concentration of UBC7. Even such high levels were unable to support conjugation to any endogenous proteins (data not shown). Removal of the loop in UBC7 did not render it capable of supporting conjugation of ubiquitin to a broad spectrum of endogenous substrates (Fig. 4B). Thus, divergent sequences in the conserved core domain itself must play important roles in determining specificity. Indeed, a few as four amino acid substitutions scattered across a large surface of the E2 core domain has previously been shown to alter specificity of conjugation (40). Interestingly, although broad spectrum incorporation of ubiquitin to protein was not observed, one specific ubiquitinated band was detected in the assays with ubc7⌬loop. Thus, the loop may still play a role in specifying substrate selectivity (Fig. 4B).
Since UBC7 appeared more highly expressed in the testis, we tried to identify specific substrates in this tissue. We fractionated the extracts of testes from rats of different ages by anion exchange chromatography, and we screened the fractions for the ability to support conjugation of ubiquitin to proteins in the presence of E1 and either UBC7 or ubc7⌬loop. Despite screening fractions derived from cytosolic, nuclear, and membrane preparations of the testis, no fractions could be identified that stimulated conjugation of ubiquitin to proteins in a UBC7-dependent manner (data not shown). Screening fractions in these assays in the presence of ubiquitin aldehyde, an inhibitor of many isopeptidases, also did not reveal any substrates (data not shown). These assays assume either that the E2 can directly conjugate to a protein substrate in the fraction or that the fraction contains both a UBC7-dependent E3 and the substrate. Since it is possible that the E3 and substrate may become separated during fractionation, pools of the fractions were made and assayed together in various combinations. Since S. cerevisiae UBC7 is known to support endoplasmic reticulum-associated degradation through binding to Cue1p (41), an integral endoplasmic reticulum membrane protein, we FIG. 4. Characterization of UBC7 and ubc7⌬loop. A, ubiquitin thiol ester formation with UBC7 and ubc7⌬loop. Bacterial lysate containing UBC7 or ubc7⌬loop was incubated in the presence of E1, 125 Iubiquitin, and ATP. The thiol ester reaction products were analyzed by nonreducing SDS-PAGE and autoradiography. The bacterial lysate without E2 was used as a negative control. B, Ubc7⌬loop does not support conjugation of ubiquitin to reticulocyte proteins as does E2 14K . E1, ATP, 125 I-ubiquitin, and a rabbit reticulocyte fraction depleted of E2s but containing E3 and substrates were incubated for 10 min with bacterial lysate expressing either the wild type UBC7, ubc7⌬loop, E2 14K (all at 50 nM), or without E2 as a negative control. The products were analyzed by SDS-PAGE and autoradiography. The arrow indicates a ubiquitinated protein in the reaction containing ubc7⌬loop. C, Ubc7⌬loop can support conjugation of ubiquitin to selected testis proteins. A testis extract prepared from ϳ200-g rats was chromatographed on a MonoQ anion exchange column. Fractions were incubated for 1 h with E1, AMPPNP, and 125 I-ubiquitin, in the presence or absence (control) of bacterial lysate containing UBC7 or ubc7⌬loop each at 100 nM. Reaction products were resolved by SDS-PAGE, and ubiquitinated proteins were detected by autoradiography. Shown are reactions with fractions eluting at the indicated salt concentrations. also mixed fractions from solubilized testis membranes in these assays. However, there was still no detectable conjugation supported by UBC7 (data not shown). Thus, UBC7 would appear to have very specific substrates in the cell which may be present at low levels and not detectable by our assays. Although genetic studies have indicated that S. cerevisiae UBC7 is involved in the degradation of abnormal proteins in the endoplasmic reticulum (24), it is quite likely that under normal conditions the number of substrates using this degradative system is limited. To date, the only known physiological substrate of S. cerevisiae UBC7 is the MAT␣2 repressor protein in yeast (22). In contrast to the absence of detectable UBC7 substrates, when ubc7⌬loop was used in these assays, incorporation of 125 I-ubiquitin into proteins was observed in fractions eluting at approximately 0.35 M NaCl following MonoQ chromatography of the cytosolic fraction (Fig. 4C). Thus, although the loop is not responsible for the inability of UBC7 to support broad spectrum conjugation to endogenous proteins, it can play a role in more subtle substrate selectivity.
Characterization of E2 17K and E2 17K ϩLoop-To evaluate more carefully the role of the loop, we inserted the 13 extra residues into E2 14K to form E2 14K ϩloop, and we determined the effects of this insertion. To test if E2 14k ϩloop can accept ubiquitin from E1, thiol ester assays were performed in vitro.
In the presence of 125 I-ubiquitin and purified E1, bacterial lysate expressing wild type E2 14K could form a thiol ester complex with 125 I-ubiquitin, but lysates expressing E2 14k ϩloop could not (Fig. 5A). This was not due to lack of expression because Western blot analysis showed that the E2 14k ϩloop was actually expressed at twice the level of E2 14K . The inability to accept ubiquitin from E1 raised the possibility that E2 14k ϩloop did not fold properly. Recently, we found that E1 can activate the ubiquitin homologue NEDD8 which is 60% amino acid identical and 80% similar to ubiquitin, but binds it more weakly than ubiquitin (apparent K m NEDD8 Ϸ25 M). 3 However, we found that both E2 14K and E2 14k ϩloop could form thiol ester complexes with NEDD8 in the presence of 125 I-NEDD8 and purified E1 (Fig. 5A). Thus the inability of the E2 14k ϩloop to accept 125 I-ubiquitin was not due to a major folding defect and UBC7 that like the loop can affect interaction of the E2 with E1 charged with ubiquitin.
Since the loop did have some effects on the ability of UBC7 to recognize substrate, we tested whether the presence of the loop affected the ability of E2 14K to support conjugation to substrates. Since E2 14K ϩloop could form a thiol ester with NEDD8, we tested whether it can support NEDD8 conjugation to endogenous proteins in the rabbit reticulocyte extract. Na-tive E2 14K can efficiently support incorporation of NEDD8 into a wide spectrum of proteins. However, E2 14K ϩloop showed a major impairment of ability to support conjugation to reticulocyte proteins (8-fold decrease, Fig. 5B). Thus, this also confirmed that the loop can play a role in substrate selectivity.
In summary, we have identified the molecular nature of rabbit reticulocyte E2 17K , the last of the major reticulocyte E2s to be so characterized. Previous workers (37) have hypothesized that hypervariable regions in E2 primary sequences may play an important role in specifying function with respect to substrate targeting. We have tested this hypothesis on the largest hypervariable region, and we have found that it does not have such a simple, well defined function. The loop residing in this hypervariable region of the E2s can indeed affect substrate selectivity of conjugation, but it can also alter the ability 3 S. S. Wing, unpublished data.  5. Characterization of E2 17K and E2 17K ؉loop. A, E2 14K ϩloop forms thiol ester with NEDD8 but not with ubiquitin. Thiol ester formation between E2 14K or E2 14K ϩloop and ubiquitin or NEDD8 were detected by incubation with E1 (100 nM), ATP, bacterial lysate containing the E2s (100 nM), and either 125 I-ubiquitin (5 M) or 125 I-NEDD8 (100 M). The bacterial lysate without any E2 was used as a negative control. After incubation, one set of reactions was quenched with Laemmli sample buffer containing 2-mercaptoethanol (2-Me) to confirm that signals detected represented thiol ester products. The reaction products were resolved by SDS-PAGE, and E2 thiol esters were detected by autoradiography. To quantitate the E2s present in the thiol ester reactions, identical amounts of bacterial lysate were analyzed by Western blot with anti-E2 14K antibody. B, addition of the loop impairs the ability of E2 14K to support NEDD8 conjugation to reticulocyte proteins. The mixture of E1, ATP, 125 I-NEDD8, and reticulocyte fraction was incubated with E2 14K or E2 14K ϩloop (50 nM) containing bacterial lysate. At the indicated times, aliquots of the reactions were removed and quenched with Laemmli buffer and 2-mercaptoethanol. The products were boiled in the presence of 2-mercaptoethanol and then analyzed by SDS-PAGE and detected by autoradiography. of the E2 to accept ubiquitin from E1. This latter feature could play a role in regulating substrate conjugation by rendering more favorable the interaction of some E2s with E1 charged with ubiquitin or possibly other ubiquitin-like molecules. More importantly, though, the effects of the loop on substrate conjugation and on interaction with E1 differed in UBC7 and E2 14K indicating that the precise effects were dependent on the nature of the core domain where the loop was attached. Thus, as was shown for the C-terminal extension of E2 25K , the function of this additional segment is dependent on sequences in the core (42). Previous studies showing that transfer of the Cterminal extension of CDC34 onto the core domain of RAD6 can confer CDC34 functions to the chimeric molecule (25,43) would suggest that specificity of conjugation can be ascribed to these additional sequences and that catalytic functions can be ascribed to the core domain. However, our findings propose a more complex model in which specificity of function is determined together by both variable sequences in the core and the additional sequence elements.