The Arabidopsis thaliana UBC7/13/14 genes encode a family of multiubiquitin chain-forming E2 enzymes.

Covalent modification of proteins by attachment of multiubiquitin chains serves as an essential signal for selective protein degradation in eukaryotes. The specificity of ubiquitin-protein conjugation is controlled in part by a diverse group of ubiquitin-conjugating enzymes (E2s or UBCs). We have previously reported that the product of the wheat TaUBC7 gene recognizes ubiquitin as a substrate for ubiquitination in vitro, catalyzing the condensation of free ubiquitin into multiubiquitin chains linked via lysine 48 (van Nocker, S., and Vierstra, R. D.(1991) Proc. Natl. Acad. Sci. U. S. A. 88, 10297-10301). Based on this activity, this E2 may play a central role in the ubiquitin proteolytic pathway by assembling chains in vivo. Here, we describe the cloning and characterization of a three-member gene family from Arabidopsis thaliana (designated AtUBC7/13/14) encoding structural homologs of TaUBC7. Like TaUBC7, recombinant AtUBC7/13/14 proteins formed multiubiquitin chains in vitro. AtUBC7/13/14 mRNAs were found in all tissues examined, and unlike related UBCs from yeast, the levels of mRNA were not elevated by heat stress or cadmium exposure. Transgenic Arabidopsis were engineered to express increased levels of active AtUBC7, for the first time altering the level of an E2 in a higher eukaryote. Plants expressing high levels of AtUBC7 exhibited no phenotypic abnormalities and were not noticeably enriched in multiubiquitinated conjugates. These findings indicate that the in vivo synthesis of multiubiquitin chains is not rate-limited by the abundance of AtUBC7 and/or involves other, yet undefined components.

A variety of essential processes in eukaryotes are regulated by selective protein breakdown via the ubiquitin-dependent proteolytic pathway (Hershko and Ciechanover, 1992;Vierstra, 1993;Ciechanover, 1994). In this pathway, protein targets are first modified by attachment of a chain of ubiquitin monomers, internally linked through Lys 48 of one ubiquitin and the carboxyl-terminal Gly 76 of the adjacent ubiquitin. This modifica-tion serves as a signal for the subsequent breakdown of the target protein by the 26 S proteasome complex (Hershko and Ciechanover, 1992). A 26 S proteasome subunit which recognizes multiubiquitin chains has been described recently and appears to have a high affinity for chains containing four or more ubiquitins (Deveraux et al., 1994;van Nocker et al., 1996). Substrates of the ubiquitin-dependent proteolytic pathway include aberrant polypeptides and important cellular regulators such as phytochrome A (Jabben et al., 1989), cyclins (Glotzer et al., 1991), p53 and c-Jun oncoproteins (Scheffner et al., 1993;Treier et al., 1994), the yeast MAT␣2 transcriptional regulator and G␣ protein Gpa1 (Hochstrasser et al., 1991;Madura and Varshavsky, 1994), and components of the NF-B transcriptional complex (Palombella et al., 1994).
Ubiquitin conjugation is an ATP-dependent, multi-step process requiring the sequential action of at least two enzymes (Hershko and Ciechanover, 1992). Ubiquitin-conjugating enzymes (E2s) 1 function by accepting activated ubiquitin from ubiquitin activating enzymes (E1s) and transferring it to a target protein. This transfer involves the formation of an ubiquitin-E2 thiol-ester intermediate where ubiquitin is linked through Gly 76 to a specific cysteine within E2. The bound ubiquitin then becomes conjugated to a target protein via an isopeptide bond between Gly 76 of ubiquitin and free lysl ⑀-amino groups within the target. In at least some cases, additional recognition factors, termed ubiquitin-protein ligases (E3s), are required for substrate recognition and ubiquitin transfer (Ciechanover, 1994;Scheffner et al., 1995).
Proteins destined for degradation via the ubiquitin-dependent proteolytic pathway become multiubiquitinated by the attachment of a chain of ubiquitin monomers linked through Gly 76 3 Lys 48 (Chau et al., 1989). The mechanism by which this intermolecular linkage is formed in vivo is unknown, but at least two possibilities exist. In a two-or multi-step process, a single ubiquitin is attached to a target protein; this ubiquitin would then serve as the site for appending additional ubiquitins. Alternatively, multiubiquitin chains could be preformed by concatenation of free ubiquitin monomers; the resulting chain would then be attached to the target in a single step (Chen and Pickart, 1990;van Nocker and Vierstra, 1993). In support of the latter possibility are the observations that (i) several types of E2s exist that catalyze the formation of free multiubiquitin chains, at least in vitro; (ii) free chains can interact with E1 and various E2s and be transferred en masse to targets in vitro with kinetics similar to ubiquitin monomers; and (iii) large cellular pools of free multiubiquitin chains exist in a variety of eukaryotes (Chen and Pickart, 1990;Vierstra, 1991, 1993).
We have previously reported the cloning of a cDNA from  (BIR 92-20331). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) U33757 (AtUBC7), U33758 (AtUBC13), and U33759 (AtUBC14).
‡ To whom correspondences should be addressed. Tel.: 608-262-8215; Fax: 608-262-4743; E-mail: vierstra@macc.wisc.edu. wheat encoding an E2, TaUBC7, that is capable of forming free multiubiquitin chains in vitro (van Nocker and Vierstra, 1991). In an attempt to better elucidate the role of E2s such as TaUBC7 in multiubiquitin chain formation in vivo, we have cloned the corresponding genes from the plant Arabidopsis thaliana, which is better suited for genetic manipulations. Here, we report the isolation and characterization of a threemember gene family (designated AtUBC7/13/14) encoding E2 proteins with 71-76% identity to TaUBC7. Like TaUBC7, these E2s are capable of synthesizing multiubiquitin chains from free ubiquitin in vitro. Transgenic Arabidopsis were engineered to express high levels of AtUBC7. This ectopic expression did not phenotypically alter the plants nor increase the pool of multiubiquitin chains. From this, we conclude that multiubiquitin chain formation in vivo is not limited by AtUBC7/13/14 enzyme levels.

Isolation of Clones for Members of the Arabidopsis AtUBC7/13/14
Gene Family-Unless otherwise indicated, A. thaliana L., ecotype Columbia was used as the source of all DNA and RNA. A 577-bp EcoRI/ ScaI fragment from the wheat cDNA, TaUBC7 (van Nocker and Vier stra, 1991), was used to probe an amplified -ZAP (Stratagene, La Jolla, CA) cDNA library prepared with poly(A) ϩ RNA isolated from green leaves . The probe was radiolabeled with 32 P-labeled dCTP by the random priming method (Feinberg and Vogelstein, 1983) and hybridized to membrane-bound DNA as suggested by the manufacturer (Zeta-Probe; Bio-Rad). Phage corresponding to positive plaques were rescued to phagemids in Escherichia coli strain XL-1 Blue MRFЈ (Stratagene). This screen resulted in the identification of the cDNA AtUBC7.
To isolate the AtUBC7-like cDNAs AtUBC13 and AtUBC14, we screened an amplified, -ZAP II cDNA library prepared with 0.5-1kilobase pair poly(A) ϩ RNA isolated from 3-day-old etiolated, hypocotyl/ cotyledon tissue pretreated with ethylene (Schindler et al., 1992). This screen utilized as the probe a 484-bp BglII fragment from the AtUBC7 cDNA. Genomic clones of AtUBC7-like sequences were isolated from an -EMBL 3 genomic library (gift of N. Crawford, University of California, San Diego) by screening with probes derived from each of the AtUBC7/13/14 cDNAs. DNA from phage corresponding to positive plaques was isolated according to the method of Sambrook et al. (1989). In all cases, DNA sequence was determined from both strands (Vieira and Messing, 1987). Nucleotide and amino acid sequence analyses employed programs of the University of Wisconsin Genetics Computer Group (UWGCG) Software Package (Devereux et al., 1984).
DNA and RNA Gel Blot Analyses-Arabidopsis genomic DNA and total RNA were isolated using Qiagen nucleic acid purification columns as suggested by the manufacturer (Qiagen, Inc., Chatsworth, CA). DNA was digested with appropriate restriction endonucleases and fractionated by electrophoresis on 1% agarose gels. To identify all potential AtUBC7-related sequences, the 484-bp BglII coding region fragment of the AtUBC7 cDNA was used as a probe. To identify specific members of the AtUBC7/13/14 family, the following probes, derived from 3Ј nontranslated or non-transcribed sequence, were used: AtUBC7, the 266-bp BclI/EcoRI fragment from the AtUBC7 cDNA; AtUBC13, the 654-bp BclI/EcoRI fragment from the AtUBC13 genomic clone; AtUBC14, the 196-bp XbaI/HindIII fragment from the AtUBC14 cDNA. The 484-bp BglII coding region fragment of the AtUBC7 cDNA was used as a probe in RNA gel blot analysis. RNA was fractionated on 1% agarose/formaldehyde gels. Nucleic acids were transferred to Zeta-Probe membranes according to Sambrook et al. (1989). Probes were labeled with 32 Plabeled dCTP via random priming (Sambrook et al., 1989) and hybridization was at 65°C for 8 -16 h in 0.25 M Na 2 HPO 4 (pH 7.2, 22°C) containing 1 mM Na 4 EDTA and 7% SDS.
RT-PCR Analyses-Oligo(dT) 15 (300 ng) was annealed to 5 g of total Arabidopsis RNA at 65°C for 5 min in a volume of 36 l of H 2 O. Reactions were then brought to 22°C over the course of 10 min. Firststrand cDNA was synthesized for 1 h at 37°C in a 50-l reaction containing the annealing reaction and final concentrations of 50 mM Tris-HCl (pH 8.3, 37°C), 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, 1 mM each dNTP, 40 units of RNase inhibitor (Promega Corp., Madison, WI), and 50 units of Moloney murine leukemia virus reverse transcriptase (Promega). The reaction was then heated to 90°C for 5 min and cooled on ice. Amplification reactions (50 l) contained up to 2.5 l of cDNA reaction, 10 mM Tris-HCl (pH 9.0, 25°C), 50 mM KCl, 0.1% Triton X-100, 2.5 mM MgCl 2 , 0.2 mM dNTPs, 1.25 unit of Taq DNA polymerase (Promega), and 240 -320 nM oligonucleotide primers. Reactions were incubated at 91°C for 4 min, followed by 30 cycles of 91°C for 1 min, 54°C for 2 min, and 72°C for 2 min. The final 72°C incubation was extended by 10 min. Reaction products were analyzed by agarose gel electrophoresis and ethidium bromide staining. PCR products were confirmed as amplifications from the desired sequence by electrophoretic migration and restriction endonuclease mapping (data not shown). Oligonucleotides specific for the agamous mRNA (5Ј-CCGAATCCGATCCAAGAAG-3Ј, 5Ј-CCGTGAATCAAACGGTTGAG-3Ј) were complementary to sequence corresponding to the 3Ј non-coding region and thus avoided conserved sequences encoding the MADS box domain (Yanofsky et al., 1990).
Production and Purification of Recombinant E2s-Using PCR with mutagenic primers, restriction sites were introduced into the AtUBC7/ 13/14 cDNAs at the presumed start codon (CATATG; NdeI) to allow insertion of each cDNA into the E. coli expression vector pET3a (Rosenberg et al., 1987). E. coli BL21(DE3) cells carrying the appropriate plasmid were induced in log phase by the addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside to the media and lysed 3 h later by sonication. Recombinant E2 proteins were purified from the soluble fraction by a combination of ammonium sulfate fractionation, gel filtration, and anion exchange chromatography (van Nocker and Vierstra, 1991).
Analysis of rE2 Activity-Thiol ester assays were performed as described previously (van Nocker and Vierstra, 1991) using purified, recombinant enzymes and 125 I-labeled ubiquitin. Di-ubiquitin conjugate formation from free ubiquitin was assayed as formerly described (van Nocker and Vierstra, 1993) using 1 unit of recombinant E2. One unit was defined as the amount of E2 protein necessary to form a thiol-ester complex with 1 pmol of 125 I-labeled ubiquitin in a 2-min reaction at 0°C containing 75 pmol of 125 I-labeled ubiquitin, 20 nM recombinant wheat E1 (TaUBA1; Hatfield et al., 1990), 2 mM MgATP, and 0.1 mM dithiothreitol in 50 mM Tris-HCl (pH 8.0, 4°C). The diubiquitin conjugate was quantified by ␥-counting of gel slices containing the product, following its localization by autoradiography.
Generation of Transgenic Arabidopsis-For overexpression of AtUBC7, a 500-bp NcoI/BclI fragment from the AtUBC7 cDNA containing the entire AtUBC7 coding region was ligated into the EcoRI/BamHI sites of pBluescript SKϪ (Stratagene). A 567-bp EcoRV/SacI fragment comprising the 500-bp cDNA fragment plus pBluescript sequence was then excised and ligated into the SmaI/SacI sites of pBI121 (Clontech Laboratories, Palo Alto, CA). For expression of antisense AtUBC7 mRNA, a 590-bp SacI/BclI fragment from the AtUBC7 cDNA containing all of the AtUBC7 coding region and 40 bp of AtUBC7 5Ј non-translated region was ligated directly into the SmaI/SacI sites of pBI121. A. thaliana L., ecotype C24 was transformed essentially as described by Valvekens et al. (1988). Progeny from initial transformants harboring AtUBC7 (denoted T 1 seedlings) were selected by kanamycin resistance (50 mg/liter). For each transformed line, non-segregating T 1 plants were identified and the corresponding T 2 plants were utilized for phenotypic and biochemical analyses.
Plant Growth Conditions-For biochemical analyses, wild-type and transgenic Arabidopsis were grown in soil in growth chambers at 18°C under 12-or 18-h photoperiods or constant illumination. Heat stress responsiveness of the AtUBC7/13/14 genes was assayed as described previously (Girod et al., 1993). To examine the response of plants to heavy metals, plants were grown on agar-solidified MS media containing 1% sucrose. Fourteen-day old seedlings were flooded with either water or various concentrations of CdCl 2 , ZnCl 2 , or CuCl 2 (0.1-10 M) and entire plants were harvested after 8 h.
Immunological Techniques-Crude protein extracts were prepared by extracting fresh plant tissue with homogenization buffer (50 mM Tris-HCl, 2 mM Na 4 EDTA (pH 8.0, 4°C), 14 mM 2-mercaptoethanol], with the addition of 0.5 mM phenylmethylsulfonyl fluoride (100 mM in isopropyl alcohol) and 100 nM pepstatin A just before use. Proteins were resolved by SDS-PAGE (Laemmli, 1970) and electroeluted onto Immobilon-P or -NC membranes (Millipore Corp.). Anti-ZmUBC7 serum was produced in a New Zealand White rabbit using as antigen a highly purified recombinant product of the Zea mays ZmUBC7 gene. 2 Anti-ZmUBC7 immunoglobulins were obtained by affinity purification from total serum, using recombinant protein affixed to Affi-Gel 15 (Bio-Rad). Immunoblot analyses employed either these immunoglobulins, or affinity purified anti-oat ubiquitin immunoglobulins as described (Jabben et al., 1989).
Purification of E2s from Arabidopsis-Crude protein extracts (see above) were prepared from green leaves and clarified by centrifugation at 20,000 ϫ g for 20 min. Protein precipitating between 40 and 70% ammonium sulfate was resuspended in 50 mM Tris-HCl (pH 8.0, 4°C) containing 1 mM Na 4 EDTA. The resulting extract was applied to DE52 cellulose, and E2s were eluted in a 100 -150 mM step of KCl in 50 mM Tris-HCl (pH 8.0, 4°C) containing 0.5 mM dithioerythritol. All protein purification steps were carried out at 0 -4°C.

RESULTS
Isolation and Characterization of the AtUBC7/13/14 Genes-As a first step toward elucidating the mechanism(s) used to synthesize multiubiquitin chains in Arabidopsis, we identified Arabidopsis homologs of wheat TaUBC7, an E2 capable of forming multiubiquitin chains in vitro. A fragment of the TaUBC7 cDNA was used as a probe to screen an amplified -ZAP cDNA expression library synthesized with mRNA isolated from Arabidopsis leaf tissue. Examination of 2.5 ϫ 10 5 phage yielded a new 783-bp cDNA (designated AtUBC7; Fig. 1) related to TaUBC7 but not to any previously characterized AtUBC genes (AtUBC1-6, Sullivan et al. (1994); AtUBC8 -12, Girod et al. (1993); and AtUBC15 (formerly AtUBC2-1), Bartling et al. (1993)). The AtUBC7 cDNA ended in a poly(A) tract of 40 residues, preceded within 30 bases by several potential polyadenylation consensus sequences (Dean et al., 1986). A single long open reading frame was identified that would encode a 18,722-Da protein of 166 amino acids with a predicted isoelectric point of 5.33 ( Figs. 1 and 2A). The derived amino acid sequence exhibited homology with the wheat TaUBC7 gene product (76% identity, 85% similarity (van Nocker and ) and with the yeast E2, ScUBC7 (52% identity, 66% similarity (Jungmann et al., 1993)). In contrast, the predicted protein exhibited at most only ϳ41% identity to members of the AtUBC1-3 E2 family, ϳ27% identity to the AtUBC4 -6 E2 family, ϳ38% identity to the AtUBC8 -12 E2 family, and ϳ58% identity to AtUBC15. Based on these calculations, we concluded that AtUBC7 represents a member of a distinct UBC gene family in Arabidopsis homologous to wheat TaUBC7. Neither TaUBC7 nor AtUBC7 are closely related in amino acid sequence to bovine E2-25 kDa (Fig. 2B), even though both E2 types are capable of assembling multiubiquitin chains in vivo (see below: Chen et al. (1991)).
To isolate additional AtUBC7-like cDNAs, an amplified -ZAP II A. thaliana cDNA expression library was screened as above. This screen of 1 ϫ 10 5 recombinant phage led to the isolation of six clones identical to AtUBC7 except for the length of the 5Ј and 3Ј non-translated sequence. All of these AtUBC7 clones had a 3Ј non-coding region extending beyond the site of poly(A) addition found in the original AtUBC7 clone. This screen also resulted in the isolation of two cDNAs, designated AtUBC14; they were identical except for the length of the 5Ј and 3Ј ends, and had 67% nucleotide sequence identity to AtUBC7 (Fig. 1). (After completion of this work, a cDNA identical to AtUBC14 was also identified by random sequencing of an Arabidopsis cDNA library (Genschik et al., 1994).) Following the isolation of a unique genomic clone designated AtUBC13 (below), the -ZAP II cDNA expression library was screened as above using a 652-bp BclI/EcoRI fragment from the AtUBC13 genomic clone as a probe. This screen of 1 ϫ 10 5 recombinant phage yielded three distinct cDNAs with complete nucleotide identity to AtUBC13 genomic sequence, ϳ80% identity to AtUBC7, and ϳ67% identity to AtUBC14 (Fig. 1). The AtUBC13 cDNAs differed only in the length of 5Ј and 3Ј nontranslated sequences (data not shown).
In an attempt to isolate genomic clones of AtUBC7-like sequences, a 484-bp BglII fragment containing 28 bp of 5Ј noncoding and nearly the entire coding region of AtUBC7 was used as a probe to screen a -EMBL 3 library containing A. thaliana genomic DNA. Examination of 1.5 ϫ 10 5 phage resulted in the detection of a genomic sequence encompassing the entire cDNA of AtUBC7. An additional clone, designated AtUBC13, was identified that contained AtUBC7-related sequence (above). To identify a genomic equivalent of the AtUBC14 cDNA, the Arabidopsis genomic library was screened using a 196-bp XbaI/ HindIII fragment containing AtUBC14 3Ј non-translated sequence as a probe. This screen resulted in the detection of a genomic sequence that included the entire cDNA sequence of AtUBC14.
The predicted protein products of the AtUBC13/14 genes were nearly identical to AtUBC7 in terms of both apparent molecular mass and charge (AtUBC13: M r ϭ 18,822, pI ϭ 5.34; AtUBC14: M r ϭ 18,728, pI ϭ 5.33). All three contain a conserved cysteine (residue 89 for AtUBC7 ( Fig. 2A)) at a position similar to that of the active-site cysteine in the yeast E2 RAD6, FIG. 2. Amino acid sequence relationships between the members of the AtUBC7/13/14 protein family and other E2s. A, amino acid sequence alignment of the AtUBC7/13/14 proteins with wheat TaUBC7 and yeast ScUBC7. Residues which are identical or similar to AtUBC7 are outlined in black or gray, respectively. The 13-residue insertion characteristic of the AtUBC7-type E2s is underlined. The presumed active-site cysteine (Cys-89 for AtUBC7) is marked with an arrow. Numbers refer to the amino acid position for AtUBC7. B, dendrogram of amino acid sequence relationships among the AtUBC7/13/14 protein family, bovine E2-25 kDa, and other E2s. Distance along the horizontal axis separating two sequences is proportional to the divergence between the sequences. Members of the yeast ScUBC4/5 family, the Arabidopsis AtUBC8 -12 family, and the human UBCH5a/b/c family are included as an example of closely related E2 protein families from different kingdoms. Sequences shown are Arabidopsis AtUBC7/13/14 (this work) and AtUBC8 -12 (Girod et al., 1993), wheat TaUBC7 (van Nocker and, yeast ScUBC7 (Jungmann et al., 1993), CDC34 (Goebl et al., 1988), and ScUBC4/5 (Seufert and Jentsch, 1990), human UBCH5a/b/c (Jensen et al., 1995), tomato UBC1 (K. Feussner and C. Wasternack, unpublished data), and bovine E2-25 kDa (Chen et al., 1991). wheat TaUBC4, and Arabidopsis AtUBC1 (Sung et al., 1990;Sullivan and Vierstra, 1991), implicating this cysteine in the formation of the thiol-ester bond with ubiquitin. The region surrounding this cysteine conformed to the E2 active site motif HPN(I/V)(X) 3-4 GX(I/V/L)C(I/L)X(I/V)(I/L) that is present in almost all E2s characterized to date. Additional cysteines are present at positions 17 and 158 in all three E2s, and at position 6 in AtUBC13. AtUBC7/13/14 contains a conserved 13-residue sequence (GDDPXGYELASER) carboxyl-terminal to the active site cysteine; a similar insertion is present also in wheat TaUBC7 and yeast ScUBC7 and ScUBC3 (CDC34). Pairwise comparisons of AtUBC7/13/14 revealed a percent deduced amino acid sequence identity of 86 -96%. In contrast, AtUBC7/ 13/14 exhibited at most only 58% amino acid sequence identity to the other Arabidopsis E2s that are fully characterized to date (Bartling et al., 1993;Girod et al., 1993;Sullivan et al., 1994). Because of the structural similarity between AtUBC7, -13, and -14, we propose that they constitute a functionally related family of E2s in Arabidopsis.
The complete nucleotide sequences of the AtUBC7/13/14 genomic clones were determined and their structures are depicted in Fig. 3. Five introns exist within the region corresponding to the cDNAs of AtUBC7 and -13, whereas four exist within AtUBC14. The introns in common interrupt the coding regions at identical positions and contain the consensus GT/AG intron borders, with the exception of the 5Ј-most intron of AtUBC7, which contained GC/AG borders (data not shown). AtUBC7/ 13/14 do not contain introns outside of the coding region, at least within the region corresponding to the cDNAs.
DNA gel blots of Arabidopsis DNA digested with a variety of restriction endonucleases revealed multiple fragments that hybridized to a coding region probe from AtUBC7 (Fig. 4). For each endonuclease used, three hybridizing fragments were identified, even at reduced stringency. Similar analyses using probes unique to AtUBC7, AtUBC13, or AtUBC14 showed that each probe specifically hybridized to one of the three fragments, indicating that AtUBC7/13/14 likely constitute the entire gene family in Arabidopsis (Fig. 4).
RNA gel-blot analysis of the AtUBC7/13/14 family detected ϳ950-base mRNAs in all tissues examined, including leaf, flower, silique, and stem ( Fig. 5A; data not shown). Immunoblot analyses using anti-ZmUBC7 immunoglobulins indicated that the corresponding protein(s) were also present in these tissues (data not shown). The great degree of nucleotide sequence homology between the AtUBC7/13/14 genes, minimal length and high A/T content of non-homologous regions, and relatively low mRNA levels precluded the use of gene-specific probes to detect individual expression patterns by RNA gel blot analysis. Consequently, the expression patterns of the three genes were analyzed individually using quantitative RT-PCR employing gene-specific oligonucleotide primers (Fig. 5B). cDNA was generated simultaneously from individual RNA pools to minimize sample variation, and PCR reactions for all primer pairs were performed simultaneously to allow comparison of the relative PCR product levels between RNA sources. To validate this method, we analyzed mRNA levels of the homeotic gene agamous, which previously had been shown to be expressed in flowers, but not in leaves (Yanofsky et al., 1990). This was confirmed by the greater levels of agamous PCR product derived from flower and silique RNA relative to that from leaf RNA (ϳ100-fold (Fig. 5B)). Primers specific to each of the AtUBC7/13/14 genes produced RT-PCR products of the expected size. Whereas levels of the AtUBC7 and AtUBC13 PCR products were invariant among leaf, flower, and silique tissues, increased levels of the AtUBC14 PCR product, relative to that of AtUBC7/13, were seen in leaves (Fig. 5B). We found no increase in expression of the AtUBC7/13/14 gene family following a heat stress (37°C for 2 h) sufficient to increase Arabidopsis HSP70 and HSP100 mRNA levels (data not shown). (In yeast, the ScUBC4/5 gene family is strongly responsive to heat stress (Seufert and Jentsch, 1990).) Unlike yeast ScUBC5 and ScUBC7 (Jungmann et al., 1993), the AtUBC7/13/14 mRNAs did not accumulate following exposure of plants to levels of cadmium (10 M) that severely inhibited growth (data not shown).
Analyses of Recombinant AtUBC7/13/14 Proteins-The biochemical properties of AtUBC7/13/14 proteins were examined following expression of the respective cDNAs in E. coli . Analysis by SDS-PAGE and immunoblotting revealed that the recombinant proteins migrated at the expected molecular mass of ϳ19 kDa and were recognized by anti-ZmUBC7 immunoglobulins ( Fig. 6A; data not shown). The recombinant E2s were enzymatically active as judged by their ability to form an adduct with 125 I-labeled ubiquitin in the presence of ATP and E1 (Fig. 6B). These adducts were unstable under reducing conditions, consistent with the linkage of ubiquitin to the E2s via a thiol-ester bond (data not shown).
Like TaUBC7, the AtUBC7/13/14 E2s conjugated ubiquitin to ubiquitin in vitro, and following prolonged incubation could generate multiubiquitin chains containing as many as seven monomers (see Fig. 8B for the SDS-PAGE profile of chains assembled in vitro by AtUBC7). To compare accurately the kinetics of these three E2s in multiubiquitin chain formation, we assayed for the assembly of di-ubiquitin from free ubiquitin using equivalent amounts of E2 activity (as determined by thiol-ester assay) (Fig. 7). AtUBC14 was the most active (1.9 mol of di-ubiquitin/min), followed by AtUBC7 (1.2 mol/min) and AtUBC13 (0.7 mol/min). These rates, however, were significantly lower than that of TaUBC7 under the same reaction conditions (3.8 mol/min).
Production and Analyses of Transgenic Arabidopsis-To understand better the in vivo functions of the AtUBC7/13/14 E2 family, we attempted to alter the levels of AtUBC7 in transgenic Arabidopsis using either sense or antisense approaches (van der Krol et al., 1990;Matzke and Matzke, 1995). If the AtUBC7/13/14 family is a limiting factor in multiubiquitin chain synthesis, we expected that an increase in active AtUBC7 would subsequently increase the ratio of multiubiquitin chains to ubiquitin monomers whereas a decrease in active AtUBC7 would decrease the ratio. The coding region from the AtUBC7 cDNA was placed in the forward and reverse orientations relative to the cauliflower mosaic virus 35 S promoter and nopaline synthase transcriptional terminator and introduced into Arabidopsis cv. C24. The cauliflower mosaic virus 35 S promoter affords high level, constitutive expression of many genes in plant tissues (Jefferson et al., 1987), whereas the nopaline synthase terminator effectively terminates transcription by plant RNA polymerases (Benfey et al., 1990). We obtained six independent transgenic lines expressing levels of antisense AtUBC7 mRNA that far exceeded the levels of the wild-type AtUBC7/13/14 mRNA (data not shown). However, the amount of AtUBC7/13/14 protein in each of these lines was not noticeably altered, and the transgenic plants were phenotypically normal by all parameters assayed. These included sensitivity to prolonged drought, water stress, heat stress, light stress, days and leaf number before flowering under short and long photoperiods, and fresh weight at time of flowering (data not shown).
With the sense AtUBC7 vector, we obtained 11 independent lines producing increased levels of AtUBC7 mRNA and protein ( Fig. 8A and data not shown). None of these lines exhibited co-suppression of the wild-type AtUBC7/13/14 genes which can occur when endogenous genes are ectopically expressed in Arabidopsis (Matzke and Matzke, 1995). AtUBC7 protein from these lines comigrated with AtUBC7 produced in E. coli and was recognized by anti-ZmUBC7 immunoglobulins (Fig. 8A). Quantitative immunoblot analysis using purified, recombinant AtUBC7 as the standard indicated that plants homozygous for the introduced DNA accumulated Ն7-fold more AtUBC7 protein than non-transformed plants. This increase was evident in almost all tissues examined, including roots, stems, leaves, inflorescence meristems, flowers, siliques, and etiolated seedlings ( Fig. 8A; data not shown). The only exception was seeds where little accumulation of AtUBC7 was detected.
To demonstrate that the AtUBC7 protein expressed in planta was enzymatically active, we partially purified AtUBC7 from transgenic leaf tissue and examined its ability to form a thiol-ester adduct with 125 I-labeled ubiquitin. Whereas the amount of adduct formed by endogenous AtUBC7 in non-transformed plants was below our limits of detection, in transgenic lines producing high levels of AtUBC7 protein we detected a ubiquitinated product that comigrated with authentic AtUBC7-ubiquitin thiol-ester adduct (Fig. 8C). The specific activity of AtUBC7 expressed in the transgenic plants was equal to, or slightly greater than, recombinant AtUBC7 produced in E. coli, as determined by quantitative thiol-ester assays (data not shown).
To test the hypothesis that AtUBC7/13/14 function in vivo to assemble multiubiquitin chains, we compared the ubiquitin conjugate profiles in cell lysates from non-transformed plants to those from AtUBC7-overexpressing plants by SDS-PAGE and immunoblotting. In wild-type Arabidopsis extracts, free multiubiquitin chains are the most abundant ubiquitin conjugates, easily detected as an immunoreactive ladder differing by ϳ8-kDa increments . As seen in Fig. 8B, free multiubiquitin chains containing from 2 to as many as 5 monomers could be detected that comigrated with authentic Gly 76 3 Lys 48 chains synthesized in vitro using AtUBC7. The chains were abundant in all tissues examined, including root, seed, leaf, stem, flower, and etiolated seedling (Fig. 8B). Surprisingly, in AtUBC7 overexpressing plants, no change was apparent in either the abundance or distribution of free multiubiquitin chains or in the abundance of other ubi-quitinated proteins relative to that of the free ubiquitin monomers (Fig. 8B). Transgenic plants producing high levels of active AtUBC7 protein also were phenotypically indistinguishable from non-transgenic plants by all assayed parameters (see above).

DISCUSSION
Multiubiquitin chains linked through Gly 76 3 Lys 48 function as a strong signal in directing proteolysis when attached to various target proteins (Chau et al., 1989;Ciechanover, 1994). Despite its importance to basic cell biology, the mechanism by which target proteins become multiubiquitinated is not yet known. We have previously reported the isolation and molecular characterization of TaUBC7, a ubiquitin-conjugating enzyme from wheat capable of forming multiubiquitin chains linked exclusively through Gly 76 3 Lys 48 from free ubiquitin in vitro (van Nocker and Vierstra, 1991). Here, we report the isolation and characterization of both cDNA and genomic clones comprising a family of three transcribed genes from Arabidopsis, designated AtUBC7, -13, and -14, encoding structural homologs of TaUBC7 that are also capable of multiubiquitin chain formation in vitro. All three recombinant Arabidopsis E2s exhibited slightly different kinetics in this reaction that were significantly lower than that of TaUBC7. However, because our studies employed a recombinant wheat E1, the lower rate of chain formation by the Arabidopsis E2s compared with TaUBC7 may reflect a preference for interacting with a more closely related E1.
DNA gel blot analysis under conditions of low stringency and extensive screenings of both genomic and cDNA expression libraries indicate that AtUBC7/13/14 constitute the entire gene family in Arabidopsis. This genomic organization is typical of that seen for other Arabidopsis E2s; the AtUBC1 and AtUBC4 families contain three members each (Sullivan et al., 1994), and the AtUBC8 family contains at least five members (Girod et al., 1993). Chromatographic analyses of E2s from wheat suggest that multiple isoforms of TaUBC7 likely exist in this species as well, but whether they are encoded by different genes has not been determined (van Nocker and Vierstra, 1991).
The AtUBC7/13/14 cDNAs were predicted to be nearly fulllength by comparison with the mRNA lengths of ϳ950 bases. For each cDNA isolated, heterogeneity was observed in the length of both the 5Ј and 3Ј non-translated regions. Some, or all, of the variability in 3Ј sequence length was an artifact of library construction. 3 However, at least for AtUBC7, the presence of two poly(A) addition sites was evident. Both RNA gel blot analysis and RT-PCR indicate that AtUBC7/13/14 are actively transcribed in most, if not all, Arabidopsis tissues. In leaves, flowers, siliques, and stems, AtUBC7/13/14 mRNAs accumulate to approximately the same abundance and are present at the same ratio, with the exception of AtUBC14, which is relatively more abundant in leaves. The AtUBC7/ 13/14 genes are also expressed in etiolated seedlings, as this was the source of the tissue from which the cDNA libraries were created. The wide tissue distribution of AtUBC7/13/14 proteins indicate that this family of E2s has a pervasive role in basic cellular functions. The corresponding mRNAs do not accumulate in response to heat stress or cadmium exposure. This data suggest that the corresponding E2s are either not involved in the response to these stresses, or are already at sufficient levels to carry out a stress-related function.
AtUBC7/13/14 encode acidic, ϳ19-kDa proteins of 166 -167 amino acids with 76% sequence identity to wheat TaUBC7 and 50% identity to yeast ScUBC7. They contain the ϳ150-amino 3 J. Ecker, personal communication. acid core domain identified in all E2s characterized to date that includes an active-site cysteine for ubiquitin transfer. In AtUBC7/13/14, the core domain is interrupted by a short (ϳ13 amino acids) insertion near the active site. Among 27 other E2s characterized to date at the molecular level, similar internal insertions are found only in yeast ScUBC3 (CDC34) and ScUBC7, and TaUBC7 from wheat (Goebl et al., 1988;van Nocker and Vierstra, 1991;Jungmann et al., 1993). Comparison of the amino acid sequence of these proteins with that of AtUBC1 and yeast ScUBC4, whose crystallographic structures have been determined (Cook et al., 1993), suggests that this internal sequence extends outward as a loop from the surface of the protein and may not be involved in intramolecular interactions. This loop may provide the necessary binding site for ubiquitin recognition during chain formation. AtUBC7/13/14 proteins show the greatest structural dissimilarity to yeast ScUBC7 and wheat TaUBC7 at the amino-terminal region. This amino-terminal heterogeneity is typical of that found among E2s from Arabidopsis and other organisms. Based on the crystal structure of AtUBC1 and ScUBC4, the extreme amino terminus likely extends away from the core of the protein (Cook et al., 1993) and may interact with E1 based on amino-terminal deletion analyses (Sullivan and Vierstra, 1991).
Given the strong amino acid conservation between the AtUBC7/13/14 proteins and ScUBC7, it is tempting to speculate that these proteins are functionally analogous, in spite of the fact that AtUBC7/13/14 mRNAs do not accumulate following exposure of the plant to cadmium. However, inferring function based on structural conservation is complicated by the presence of at least one other yeast E2 exhibiting homology to AtUBC7/13/14, the cell cycle regulator CDC34. Although CDC34 differs in overall structure (containing a negatively charged domain appended to the carboxyl terminus of the core), within the core region CDC34 exhibits as great an amino acid sequence similarity (68%) to AtUBC7 as does ScUBC7 (66%).
Ectopic expression of AtUBC7 mRNA in transgenic Arabidopsis resulted in the accumulation of high levels of active protein in most tissues examined. However, no change in ubiquitin conjugate profiles were seen, nor were any phenotypic effects evident. We offer several explanations to account for this. First, AtUBC7 may function in vivo in capacities other than in multiubiquitin chain formation. Second, it is possible that the AtUBC7/13/14 family comprises only one of several E2 isoforms that contribute to the cellular pool of multiubiquitin chains in Arabidopsis. For example, in addition to TaUBC7 and AtUBC7/13/14, the ability to synthesize multiubiquitin chains in vitro is common to at least one other E2, bovine E2-25 kDa. However, the plant and bovine E2s are not structural homologs as they differ in size and share less than 34% overall amino acid identity (Fig. 2B). A potential plant counterpart of E2-25 kDa has been identified from a tomato cDNA library (exhibiting ϳ48% identity to E2-25 kDa; Fig. 2B), but whether it is also capable of making multiubiquitin chains is unknown. Thus, an increase in AtUBC7 level might not lead to a proportional increase in multiubiquitin chain levels. Third, the formation of multiubiquitin chains may not be limited by the enzyme(s) required for ubiquitin polymerization, but by available quantities of free ubiquitin monomers. Fourth, other enzymes (e.g. other E2s and E3s) may be required for chain formation in vivo in addition to the AtUBC7/13/14 and may be limiting. It is tempting to speculate that the members of the AtUBC7/13/14 family may act in concert or with other E2s/E3s to promote chain formation. In support of this, it appears that AtUBC7 physically interacts with AtUBC13 and AtUBC14 in vivo, 4 and evidence is accumulating for the concerted action of multiple yeast E2s in the multiubiquitination and degradation of MAT␣2 (Chen et al., 1993).
Future studies of the role of the AtUBC7/13/14 family in multiubiquitination will depend on creating or identifying mutants deficient in these E2s in Arabidopsis. Although high level expression in transgenic Arabidopsis can often result in cosuppression of the endogenous genes (Matzke and Matzke, 1995), we detected no suppression of the endogenous AtUBC7/ 13/14 genes in any transgenic line. Nor were we able to reduce the levels of AtUBC7 protein through high level expression of antisense mRNAs. An alternative approach would be to interfere with the function of AtUBC7/13/14 E2s by expressing structurally altered forms. A particularly attractive possibility in this regard would be a Cys 3 Ser substitution at the active site. This substitution in yeast RAD6 and AtUBC1 still allows ubiquitin to bind to the E2s, in this case through an ester linkage, but precluded transfer of the bound ubiquitin to substrates (Sung et al., 1990;Sullivan and Vierstra, 1993). Bailly et al. (1994) have shown that such inactive E2s can act in a dominant negative manner when expressed to high levels in yeast.