Mouse ODC (ornithine decarboxylase) is quickly degraded by the 26S proteasome in mammalian and fungal cells. Its degradation is independent of ubiquitin but requires a degradation signal composed of residues 425–461 at the ODC C-terminus, cODC (the last 37 amino acids of the ODC C-terminus). Mutational analysis of cODC revealed the presence of two essential elements in the degradation signal. The first consists of cysteine and alanine at residues 441 and 442 respectively. The second element is the C-terminus distal to residue 442; it has little or no sequence specificity, but is intolerant of insertions or deletions that alter its span. Reducing conditions, which preclude all well-characterized chemical reactions of the Cys441 thiol, are essential for in vitro degradation. These experiments imply that the degradative function of Cys441 does not involve its participation in chemical reaction; it, instead, functions within a structural element for recognition by the 26S proteasome.
The proteasome plays a central role in intracellular protein degradation. Most of its substrates are recognized via polyubiquitin chains . In contrast, the enzyme ODC (ornithine decarboxylase) is recognized by the proteasome through a mechanism independent of ubiquitination . Mouse ODC contains a degradation signal composed of its C-terminal 37 amino acids [cODC (the last 37 amino acids of the ODC C-terminus)] . This domain, which is conserved in vertebrates (Figure 1), acts as a portable degradation tag when added to the C-terminus of various proteins, such as GFP (green fluorescent protein). The cODC degron also functions when expressed exogenously in the cells of non-vertebrate species, including fungi  and plants . Mutation of ODC Cys441, within cODC, was found decades ago to create stabilizing alleles . The mutation of Cys441 also stabilizes ODC protein expressed in yeast . Thus it is apparent that there is a conserved mechanism by which the 26S proteasome recognizes ODC through Cys441. However, the biochemical function of Cys441 in degradation and the specific sequence context and conditions required for this function remain unclear. In our previous work, we showed that removing the Cys441 thiol by mutation (C441A) or replacing it with a hydroxyl (C441S) impairs an early step of degradation, recognition by the 26S proteasome [7,8]. One possible role of Cys441 in facilitating interaction with the 26S proteasome may be structural. It may simply be used as a recognition element, in a manner that does not depend on its chemical modification. Alternatively, proper cODC function may require chemical alteration of the thiol group. We show here that reducing conditions are necessary for degradation of substrates that are dependent on cODC, but not for substrates that utilize polyubiquitin as a degradation signal. Mutation of the adjacent Ala442 also impaired function, implying that residues Cys441 and Ala442 each contribute to recognition. Truncation and deletions within the C-terminal end of cODC distal to Cys441-Ala442 showed that moving the C-terminus closer to or further from this pair also impaired degradative function. Although the length of the free C-terminus is constrained, its sequence is not.
Alignment of vertebrate cODC sequences
Strain and plasmids
The yeast strain MHY501 (his3-D200, leu2-3,112, lys2-801, trp1-1, ura3-52, MATα) was used as the wild-type for the present study. Point mutations in ODC were introduced by overlap PCR. The coding region of ODC was amplified and cloned on to p416ADH1 yeast expression vector . The amino acid sequence of cODCC441A random is -EFHGPNLPVAAMVTIADEEAMDRSPAGDQASCPERESQ-. The last ten amino acids of above sequence were inserted between 450 and 451 to make ODC451::10, and SCPER was inserted to make ODC451::5.
Recombinant protein expression
The expression vector for ODC (Mus musculus) and AZ1 (antizyme 1) (Rattus norvegicus) was pQE30 (Qiagen) with a His6 tag at the N-terminus as described previously . DHFR–cODC [the human DHFR (dihydrofolate reductase) fused to cODC] and DHFR were cloned into the pQE30 vector and expressed in Escherichia coli . Site-directed mutagenesis was carried out using the megaprimer method . All PCR products used for in vitro translation were verified by sequencing.
ODC degradation assay in reticulocyte lysate
The site-directed mutated ODC was in vitro transcribed and translated in reticulocyte lysate (TNT; Promega, Madison, WI, U.S.A.). All the PCR products used for in vitro translation contained a T7 promoter and Met-Gly-His6 positioned upstream of the ODC second amino acid. AZ1-stimulated ODC degradation was conducted in reticulocyte lysate . Briefly, 35S-labelled ODC was prepared by in vitro translation and incubated with recombinant AZ1 or in the absence of AZ1 [in a buffer containing 50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20 and 1 mM DTT (dithiothreitol)] and an ATP-regenerating system at 37 °C for 1 h. The reaction was stopped by the addition of SDS/PAGE loading buffer, followed by SDS/PAGE analysis. The autoradiogram was scanned and quantified with TotalLab software from Nonlinear Dynamics. ATP-regenerating buffer contained 30 mM Tris/HCl (pH 7.5), 5 mM MgCl2, 2 mM DTT, 1 mM ATP, 10 mM phosphocreatine and 1.6 mg/ml creatine kinase final concentration. Where indicated, DTT was eliminated from the ATP-regenerating buffer. The percentage degradation of mutant forms of ODC (ODCmut) compared with that of wild-type ODC (ODCwt) (Figure 2) is calculated as 100×(1–ODCmut+AZ/ODCmut–AZ)/(1–ODCwt+AZ/ODCwt–AZ), where, for example, ODCmut+AZ is the intensity of the mutant ODC autoradiographic band after incubation with AZ1.
Alanine scanning mutagenesis within the sequence surrounding Cys441
Protein degradation using purified components
Purified [35S]methionine-labelled recombinant His6–TEV (tobacco etch virus)–FLAG–ODC was made and degraded in vitro by using purified components as described previously . The percentage of ODC degradation was determined by dividing the trichloroacetic acid-soluble counts by the total input ODC counts. His6-tagged Ub5-DHFR degradation was carried out in the presence of 1.2 μM ubiquitin-aldehyde (Boston Biochem) and detected by 1:3000 dilution of anti-His6 antibody (Amersham Biosciences). Mammalian 26S proteasomes were purchased from Biomol (Plymouth Meeting, PA, U.S.A.).
Yeast manipulations followed standard protocols . Other procedures were as described previously . For pulse–chase experiments, cells in exponential growth phase were harvested, washed with SD medium (selective drop-out medium), resuspended in SD–Met containing 0.2 mCi of [35S]methionine/cysteine (NEN) and labelled for 5 min at 30 °C. Cells were harvested, resuspended in 400 μl of SD–Ura containing 10 mM methionine/cysteine and 0.5 mg/ml cycloheximide and continuously incubated at 30 °C. Samples of 100 μl were taken periodically and suspended in lysis buffer (50 mM Hepes, pH 7.5, 1% Triton X-100, 150 mM NaCl, 1 mM EDTA and 1 mM PMSF). Cells were immediately disrupted by shaking with glass beads in a Bead Beater (Biospec, Bartlesville, OK, U.S.A.) followed by centrifugation at 18000 g for 10 min. Anti-FLAG M2–agarose beads (Sigma, St. Louis, MO, U.S.A.) were added to the supernatant followed by 40 min of incubation at 4 °C. The volume of sample for immunoprecipitation was adjusted to contain equal acid-insoluble counts. The resin was then washed four times with lysis buffer and boiled with SDS/PAGE sample buffer.
Non-reducing PAGE and blotting of biotinylated proteins
To examine the effect of reducing conditions on disulfide bonds, a small molecule, Profound Mts-Atf-Biotin label transfer reagent (Pierce, Rockford, IL, U.S.A.), was conjugated with recombinant Sem1 protein, as described in the manufacturer's protocol. This reagent contains a disulfide, which is required for maintaining the association of a biotin group with the target protein. After incubation with various reducing reagents, the protein samples were mixed with SDS/PAGE sample buffer free from reducing reagent. The protein sample was then boiled for 10 min and subjected to SDS/PAGE, followed by transfer to a nitrocellulose membrane. The membrane was blocked by 7% (w/v) non-fat dried skimmed milk powder in TBST (10 mM Tris/HCl, pH 7.5, 150 mM NaCl and 0.2% Tween 20) and incubated with streptavidin–peroxidase conjugate (Roche, Penzberg, Germany). Signal was detected by an ECL® detection kit (GE Healthcare, Uppsala, Sweden).
Mutation of individual residues surrounding Cys441 does not impair cODC degron function
As C441A and C441S mutations, the most conservative available, prevented degradation of ODC, we investigated the possibility that Cys441 forms a recognition element in conjunction with other residues that lie nearby in the primary sequence. To do so, we individually mutated to alanine each non-alanine amino acid within the region encompassing residues Thr436–Met447. We tested the proteasome-mediated turnover of each mutant. Although cODC functions as an autonomous degradation tag, in the context of native ODC, its degradative activity is enhanced by the protein AZ1, which forms an AZ1:ODC heterodimer in which cODC is more accessible than in the native ODC–ODC homodimer. The AZ1-stimulated turnover of each of the above mutants was examined in a reconstituted degradation system by using rabbit reticulocyte lysate as a source of proteasomes. If Cys441 takes part in functionally significant side-chain interactions with an amino acid in this local neighbourhood, one or more of these alanine substitutions should be stabilizing. However, none of the alanine substitution mutants tested stabilized ODC, except for C441A, which profoundly reduced activity (Figure 2). We conclude that no other residue in the region examined makes side-chain interactions with Cys441 that are critical for ODC degradation or that are independently required for ODC degradation. Additionally, we tested the effect of an H450A mutation, a possible proton donor; this mutation also did not impair turnover. These investigations exclude the most plausible interactions of Cys441 with other residues within cODC.
Cys441 is intolerant of positional change
To assess whether Cys441 can function only in its precise molecular context, we moved Cys441 to each adjacent position in cODC. Since the sequence of the local region is Ser440/Cys441/Ala442 in wild-type cODC, we replaced a cysteine residue at position 440 of a C441S mutant or at position 442 of a C441A mutant, thus forming ODC with the mutations S440C/C441S and C441A/A442C. As shown in Figure 3(A), swapping the cysteine residue with either of the two adjacent residues stabilizes ODC, reducing degradation in the experiment shown from 25% to less than 5% in each case. The effect of these sequence changes was also examined by pulse–chase analysis of ODC turnover in yeast cells. Introduction of a cysteine residue at either position 440 or 442 failed to restore degradative function of Cys441 mutants (Figure 3B). Two possible mechanisms could account for the stabilization of ODC in these mutants: stabilization depends on either the loss of Cys441 or a dominant-negative effect of cysteine at position 440 or 442. To distinguish these possibilities, we reintroduced cysteine at position 441 in the above mutants to create Ser440/Cys441/Cys442 and Cys440/Cys441/Ala442. Using pulse–chase analysis in yeast cells, we found that a cysteine substitution at 440 does not impair degradation, but cysteine at 442 stabilizes ODC. This result suggested a potential role for Ala442, the single local residue not previously susceptible to analysis by alanine scanning mutagenesis. We therefore examined the A442P mutant and found it to be stable. The role of Ala442 was further studied by creating the more conservative mutations A442G and A442L. As shown in Figure 3(C), A442G was slowly degraded; in contrast, A442L was extremely stable. Thus we conclude that Ala442 cannot be changed to residues with a more bulky side chain. Taken together, the results above imply that Cys441 must be precisely positioned in the native context to act as a degradation signal and may be recognized together with the adjacent Ala442 residue. Considering the requirement of an adjacent Ala442 after Cys441, we made the Cys440/Ala441/Ala442 mutant, to determine whether Cys441 is movable when it accompanies alanine. Surprisingly, this mutant was completely stable, which suggested that fixed positioning of the Cys441/Ala442 pair is necessary for degradation (Figure 3D).
Mutation of Cys441 and adjacent amino acids
Cys441 does not form an intramolecular disulfide bond
The thiol group of cysteine acts as the catalytic centre of some enzymes, by undergoing an oxidative reaction, by acting as a proton donor or as a thiolate anion . Cys441 may act catalytically or, alternatively, Cys441 may simply be a part of a cODC structural element important for proteasome recognition. We had previously made a series of mutations intended to distinguish these molecular mechanisms. The previously described C441A mutation effectively removes the thiol of Cys441, and the C441S mutation substitutes a hydroxy group for the thiol. Among the natural amino acids, the C441S alteration best preserves the size, hydrophobicity and polarity of the cysteine residue. The stabilization of ODC by the C441S mutation  implies that the hydroxy group cannot replicate the functional properties of the thiol of Cys441. Under permissive reduction–oxidation conditions, a pair of cysteine residues can form a disulfide bond. Perhaps a disulfide bond, or a cycle of bond formation and reduction involving Cys441, is essential for cODC-mediated degradation. Mouse ODC contains 12 cysteine residues, but most can be excluded as possible disulfide partners of Cys441, because the cODC portable degradation tag contains only two cysteine residues, Cys441 and Cys454. Thus, within cODC, the possible partner for bond formation with Cys441 is restricted to Cys454. To test whether disulfide bonding between Cys441 and Cys454 is needed for degradation, we changed Cys454 to alanine and subjected the mutant to an in vitro degradation assay using reticulocyte lysate as a proteasome source. C454A did not impair degradation, eliminating the potential for disulfide bond formation within cODC to be critical for its function (Figure 4A). A pulse–chase experiment also revealed a similar result (results not shown). Another possible bonding partner of Cys441 is a cysteine thiol within a proteasome protein. We examined the AZ1-stimulated degradation of ODC with and without DTT, a reagent that can prevent or reverse disulfide bond formation. Using purified rat 26S proteasomes, we measured the production of acid-insoluble proteolysis products from 35S-labelled ODC. To our surprise, degradation was observed only under reducing conditions (Figure 4B). Next, in order to assess whether reducing conditions are generally required for proteasome function, we tested another substrate of the 26S proteasome, oligo-ubiquitylated DHFR (Ub5-DHFR), a previously described substrate whose degradation occurs through a Lys48-linked N-terminal penta-ubiquitin chain . As shown in Figure 4(C), this ubiquitin-dependent substrate was degraded equally effectively in the presence of 0.1 mM or 2 mM DTT; the lower DTT concentration is insufficient to sustain degradation of ODC. In vitro degradation of ubiquitinated substrates in the absence of reducing agents has been reported previously (e.g. ), consistent with the present result. In addition, we tested the degradation of a substrate that is tethered to the proteasome and requires neither polyubiquitin nor cODC for association . Rpn10–GFP–cODCC441A degradation by the Rpn10Δ 26S proteasome did not require reducing conditions (results not shown). Strong reducing conditions therefore favour the turnover of ODC, but this is not a general requirement for the proteolytic function of proteasomes. Taken as a whole, these results support the conclusion that ODC degradation does not require that Cys441 participate in disulfide bond formation or other chemical processes involving the oxidation of its thiol group, and indeed suggest that chemical modification of the thiol precludes its degradative function.
Effect of disrupting potential disulfides
cODC-mediated degradation requires reducing conditions
The observation that ODC turnover requires reducing conditions led us to examine whether this property also applies to proteins that are made labile by appending the cODC degradation tag. We tested this by using a DHFR–cODC construct in which the cODC tag was appended to the C-terminus of the human DHFR open reading frame . As described above, the degradation of Ub5-DHFR, which is recognized through a ubiquitin chain, is DTT-independent in vitro. However, as shown for ODC, the degradation of DHFR–cODC is also strongly promoted by reducing conditions (Figure 5A). Thus the requirement of proteasome-mediated degradation of ODC for reducing conditions is a property inherent in its C-terminal 37 amino acids. One possible role of Cys441 in cODC function is that it forms a thioester bond, which may be tolerant to 2 mM DTT. Such a putative bond is not likely to be a cyclic association with the carboxyl of an acidic residue within cODC: the alanine scanning result excluded interaction of Cys441 with the local structure of cODC; the known protein intramolecular thioesters, e.g. in complement protein C3, involve cyclization with a nearby acidic residue . However, degradation may require that Cys441 form a thioester with a protein of the 26S proteasome. We increased the concentration of DTT to 10 mM, aiming to cleave any thioester bond (or poorly accessible disulfide bond) and examined whether this condition impaired degradation of DFHR–cODC. Degradation persisted in the presence of 10 mM DTT, suggesting that thioester-bond formation between cODC and the proteasome was not required. In addition to disulfide or thioester bond cleavage, 10 mM DTT is also sufficient for reduction of cysteine residues oxidized to the sulfonic or sulfinic state. These results taken together effectively eliminated all but the simplest hypothesis: the thiol group of Cys441 must be maintained in a reduced state to act as a recognition signal for the 26S proteasome, and does not act as a bonding partner with other residues.
Comparison of biochemical conditions required for disulfide bond reduction and cODC-dependent degradation
Additionally, the structurally unrelated reducing agents 2-mercaptoethanol and TCEP [tris-(2-carboxyethyl)phosphine] also supported degradation (Figure 5B). To more readily evaluate the redox status of disulfide bonds under various experimental conditions, we utilized a protein attached to biotin through a disulfide bond. Strongly reducing conditions supported degradation and fully cleaved biotin via disulfide reduction of the model protein. In contrast, 0.1 mM DTT and 0.2 mM 2-mercaptoethanol did not support degradation of ODC and reduced the disulfide bond of the protein-linked biotin only partially (Figure 5C). Thus we conclude that the degradation demands disulfide reduction.
Spacing between Cys441 and other elements of cODC
Our previous studies revealed that cODC-mediated proteasome association and peptide entry are two separable events. Within cODC, Cys441 functions as a proteasome association element , while the C-terminal end of cODC initiates entry into the proteasome . If each of these must interact with a distinct specific site within the proteasome, changing the native spacing between these elements (20 residues intervene between Cys441 and the cODC terminus in the native primary sequence) could have functional consequences. We first asked whether increasing the spacing interfered with function. To move the end further from a functional copy of Cys441, we duplicated cODC, forming ODC with a tandem cODC–cODCC441A at its C-terminus. In this construct, the first ODC copy is wild-type and includes a wild-type Cys441, and the distal C-terminal copy of cODC carries a mutant, C441A. As controls, we also constructed and compared forms of ODC terminated by cODC–cODC, cODCC441A–cODC and cODCC441A–cODCC441A. As expected, cODCC441A–cODC was normally degraded, but cODC–cODCC441A was stable (Figure 6A). Also as expected, the duplicate wild-type copies were functional, and the duplicated mutant copies, cODCC441A-cODCC441A, were not. This result suggests that Cys441 requires a nearby free end and that it cannot collaborate with a too-distant cODC copy that is made defective in its Cys441 association element, one that is displaced from its native position by an extra 37 residues. Therefore cODC cannot function in the interior of the molecule. To rule out the possibility that cODCC441A retains a structure that confers a dominant stabilizing effect on cODC-cODCC441A, we randomized the sequence of cODCC441A, retaining its amino acid composition. The cODC–cODCC441Arandom, like cODC–cODCC441A, was stable, implying that it is excess distance rather than an inhibitory function of the distal mutant copy that precludes collaboration. To test whether a lesser displacement also precludes functional interaction, we inserted a random 10-amino-acid sequence between residues 451 and 452 of ODC. This insertion also stabilized in contrast with rapid degradation of wild-type FLAG–ODC, but a smaller insertion of five amino acids did not stabilize (Figure 6B). We next decreased the distance between the end of cODC and Cys441. Truncation of the last four residues (458–461) stabilized, as did deletion of the last five (results not shown). Both an internal deletion of the next five, residues 452–457, and that of the next five, residues 447–451, were also stabilizing (Figure 7). Because deleting each tract of five residues had similar effects, as did a ten-residue insertion, we concluded that polypeptide chain length between Cys441 and the C-terminus is strongly constrained. To investigate whether sequence as well as size matters, we randomized the last five amino acids from ARINV to VINAR. We found that this mutant, designated (457–461)random in Figure 7, is normally degraded. We also tested an inversion of the last five and the second last five amino acid sequences, shown as ‘(457–461)-(452–456)’. This mutant was again rapidly degraded. These results imply that the sequence is not the determinant of stability, but the length of the C-terminus is important. We tested the additional point mutations A457W and S456A and found that neither stabilized ODC (results not shown).
cODC functions only at the C-terminus
Deletion analysis of the C-terminal region of cODC
Substrates of the proteasome must be attracted to the proteasome and, once there, present an unstructured region that serves as an entry point and site of initial degradation . For most substrates, a polyubiquitin chain serves the first function, but there is a subset that uses an alternative degron [20,21]. The ODC degron cODC has been previously shown to subsume both the required functions within a 37-amino-acid tract . Cys441 within cODC is critical for proteasome association. In the present study, we show that the thiol group of Cys441 does not require oxidative chemical change, but instead functions within a structural motif that includes the adjacent Ala442. Mutations at the Ala442 residue adjacent to Cys441 that increase the size of the alanine side chain were found to impair function. Taken in conjunction with our previous work characterizing the function of Cys441 , our present mutagenesis data and chemical reduction data imply that Cys441 and Ala442 function together as a recognition element in which the cysteine thiol must be reduced and the side chain of the next residue must be no larger than a methyl group. We further show that spacing between the Cys441-Ala442 pair and the C-terminal end of cODC, approx. 20 residues in the native sequence, must be maintained. Their functional interaction was disturbed by moving them ten amino acids further apart or five residues closer together. However, milder changes, deleting one amino acid or inserting five amino acids, did not perturb degradation. Additionally, deletions within cODC on the N-terminal side of Cys441, performed as block deletions of five amino acids, failed to abolish its degradative effect in the context of GFP–cODC (results not shown).
Previously, it has been reported that deletion of ODC residues 447–451  in mammalian cells was not stabilizing. However, we found here that the same deletion of ODC was stabilizing in yeast cells. The tandem cODC degron experiment, as well as our earlier studies , shows that cODC does not work as a degradation signal when in a non-terminal position. These results imply that a proximal free ODC C-terminus is required for Cys441-dependent degradation.
The C-terminus of ODC has not been detected by X-ray structural analysis [24,25]. This suggests that cODC is unstructured, or assumes a structure induced by interaction with its unknown binding partner, one important for recognition by the 26S proteasome. Alternatively, the lack of electron density attributable to cODC may indicate that it has a defined structure, but is highly mobile with respect to the rest of the ODC polypeptide. The portable nature of cODC, which seemingly confers liability when attached to the C-terminus of any protein, implies that its structure, if any, can act autonomously.
Ubiquitin chains and cODC compete for recognition by proteasomes . Such studies, based on functional competition of cODC with ubiquitin chains, imply identity or overlap of their binding sites and suggest that cODC can act as a molecular mimic of ubiquitin chains. Ubiquitin-binding motifs require a hydrophobic patch on the ubiquitin surface; alanine substitutions therein abolish binding ability . However, our mutational analysis showed that individual hydrophobic residues around Cys441 can be either deleted or mutated without changing protein degradation.
Cysteine is widely involved in the ubiquitin–proteasome system. The catalytic centres of deubiquitinating enzymes, E2 ubiquitin-conjugating enzymes and E3 ubiquitin ligases all contain a cysteine residue . Some E2 enzymes are known to build ubiquitin chains on their catalytic centres through a thioester bond [27,28]. Also, ubiquitin chain formation on a cysteine residue has been reported . These chemical modifications, which depend on thioester formation, are destroyed by 10 mM DTT, thus clearly excluding a role for this structure in the function of Cys441 in ODC. A cofactor, AZ1, binds to ODC and enhances degradation, but this is not an essential event for ODC degradation in yeast cells or animal cells. Also, ODC with a C441A mutation, which is extremely stable, functionally binds to AZ1 in the yeast two-hybrid assay (J. Takeuchi, unpublished work). The Cys441 and the adjacent Ala442 are invariant among vertebrates; however, other conserved amino acids of cODC were revealed not to be important for degradation.
Ub5-DHFR was a gift from the late Professor Cecile Pickart (Johns Hopkins University, Baltimore, MA, U.S.A.). J. T. was supported by a fellowship from the Uehara Memorial Foundation. This work was supported by National Institutes of Health grants GM45335 and GM074760 to P. C.
Present address: Department of Biopharmaceutical Sciences, University of California, San Francisco, CA 94143, U.S.A.