Proteasomal Degradation of Spermidine/Spermine N 1 -Acetyltransferase Requires the Carboxyl-terminal Glutamic Acid Residues*

The rapid turnover of spermidine/spermine N 1 -acetyl-transferase (SSAT), a key enzyme in the regulation of polyamine levels, was found to be mediated via ubiquitination and the proteasomal system. SSAT degradation was blocked by the binding of polyamines or of the polyamine analog, N 1 , N 12 -bis(ethyl)spermine (BE-3-4-3), to the protein, providing a mechanism for the increase of SSAT activity in response to these agents. Site-directed mutagenesis indicated that a number of residues including arginine 19, cysteine 122, histidine 126, glutamic acid 152, arginine 155, and methionine 167 were needed for protection of SSAT by BE-3-4-3. These residues have previously been shown to reduce the affinity for the binding of polyamines to the SSAT protein, and these results indicate that the change in protein configuration brought about by this binding renders the protein resistant to proteasomal degradation. Mutations to ala-nines of residues arginine 7, cysteine 14, and lysine 141 also prevented the protection by BE-3-4-3, and these residues may be required for the formation of the pro-tected conformation. The rapid degradation of SSAT required the carboxyl-terminal region of the protein, and the two terminal glutamic acid residues at positions 170 and 171 were found to be of critical importance. Truncation of the protein to remove these residues or the mutation of either of these acidic residues to glutamine completely abolished the rapid degradation of SSAT. The addition of two extra lysine

Spermidine/spermine N 1 -acetyltransferase (SSAT), 1 which converts spermidine and spermine into their N 1 -acetyl derivatives, is an important enzyme in mammalian cells that prevents the overaccumulation of polyamines by facilitating their excretion and degradation (1)(2)(3). Alterations in SSAT activity are brought about by changes in amount of enzyme protein (4,5), and the enzyme is highly inducible by a variety of hormones, physiological stimuli, drugs, and toxic agents. It is also strongly induced by polyamines, and it has been suggested that a rise in the free polyamine content is an intermediary in the induction by other agents (6). The most potent inducers of SSAT are polyamine analogs that have substitutions on the terminal nitrogens and, therefore, are not substrates for acetylation (1,(7)(8)(9)(10)(11)(12). SSAT is very highly regulated by polyamines or polyamine analogs, and convincing evidence has been obtained for regulation at the levels of transcription (8,13), mRNA stabilization (13,14), mRNA translation (15), and protein degradation (15)(16)(17)(18).
SSAT is known to be a homodimer of a subunit containing 171 amino acids (19,20) (Fig. 1), and residues making up part of the acetyl-CoA (20,21) and polyamine (20,22) binding sites have been identified by site-directed mutagenesis. Dimerization is needed for the formation of the active site, and complementation experiments with inactive mutants have shown that the active site involves residues from both subunits (20).
It is a very interesting feature of the polyamine biosynthetic pathway that all three of the key enzymes that regulate polyamine levels, ornithine decarboxylase (ODC), S-adenosylmethionine decarboxylase, and SSAT, turn over very rapidly (6,23,24). The turnover of ODC, which is mediated by a polyamine-inducible protein termed antizyme, has been studied extensively (25)(26)(27)(28)(29), but there is no published information available on the mechanism of degradation of S-adenosylmethionine decarboxylase and SSAT.
The present studies were carried out to obtain a better understanding of the degradation of SSAT and its regulation by the natural polyamines and the polyamine analog, N 1 ,N 12bis(ethyl)spermine (BE-3-4-3). Previous studies have indicated that SSAT turns over very rapidly and that the half-life is greatly increased by polyamines or polyamine analogs (5,(15)(16)(17)(18). It is very difficult to measure the turnover of SSAT accurately in cells with low polyamine levels because the content of SSAT protein is extremely small under such noninducing conditions. However, estimates of a half-life of less than 1 h have been made (15)(16)(17), and this half-life is increased by more than an order of magnitude in the presence of BE-3-4-3 and other polyamine analogs. Experiments have been carried out in which the sensitivity of the SSAT protein and various mutants prepared by site-directed mutagenesis to degradation by proteases was used to investigate the binding of polyamine analogs to the protein and the subsequent conformational changes (22). There have been no reported studies on the mechanism of cellular degradation of SSAT or of the residues needed for the rapid turnover and for mediating the stabilization by polyamines.
In the present report we provide evidence that SSAT is a good substrate for degradation via the proteasomal/ubiquitin pathway (30 -32), that this degradation is prevented by the binding of polyamines or BE-3-4-3 to the protein, and that interaction with the proteasome requires the glutamic acid residues located at the carboxyl terminus of the protein. The similarity of the carboxyl end of the molecule to the PEST sequences known to be involved in the degradation of rapidly turning over proteins including ODC is also discussed.

EXPERIMENTAL PROCEDURES
Materials-Oligodeoxynucleotides were synthesized in the Macromolecular Core facility, Hershey Medical Center, or were purchased from Life Technologies, Inc. [␣-35 S]Thio-dATP was purchased from Amersham Corp. L-[ 35 S]Methionine (translation grade) was obtained from DuPont NEN. [1-14 C]Acetyl-CoA (50 Ci/mol) was purchased from ICN Biochemicals (Costa Mesa, CA). TNT T7 coupled reticulocyte lysate translation system and RNasin were obtained from Promega (Madison, WI). Qiagen plasmid purification columns, the pQE-30 plasmid and Qiaquick polymerase chain reaction purification and gel extraction kits were from Qiagen (Chatsworth, CA). Rabbit reticulocyte lysate prepared from phenylhydrazine-treated New Zealand White rabbits for use in degradation assays was obtained from Cocalico Biologicals (Reamstown, PA). ATP, creatine phosphokinase, phosphocreatine, 2-deoxyglucose, hexokinase, and cycloheximide were purchased from Sigma. ATP␥S was purchased from Boehringer Mannheim. Calpain inhibitor I (N-acetyl-L-leucinyl-L-leucinal-L-norleucinal) was obtained from Calbiochem. Full-length and 7.6-kDa fragments of human S5a proteasomal subunit were generously provided by Dr. Martin Rechsteiner, Department of Biochemistry, University of Utah, Salt Lake City, UT. Polyclonal anti-SSAT antibody was prepared as described previously (4). Protein A from Staphylococcus aureus Cowan strain was purchased from ICN Biochemicals.
Expression and Degradation of SSAT-The plasmid pSAT9.3 containing the SSAT cDNA in Bluescript (19) and pGEM-ODC containing the ODC cDNA (33) were used to express the respective proteins from the T7 promoter of each vector. All mutations were introduced into the SSAT cDNA using pSAT9.3 as the template for site-directed mutagenesis as described previously (20,22). The sequence of the entire coding region of all mutant cDNAs was checked to ensure the absence of secondary mutations. Plasmid DNA was purified from DH5␣ Escherichia coli extracts on Qiagen-tip 500 anion exchange columns according to the manufacturer's directions.
The 35 S-labeled wild type or mutants of SSAT (and wild type ODC) were synthesized from plasmid (0.125 g/12.5 l) in the TNT-coupled transcription/translation system as described previously (20,22). Degradation of SSAT was studied using an in vitro degradation system essentially as described for ODC (33). Briefly, the 35 S-labeled proteins generated as described above were used as substrates for degradation by incubating 4-l aliquots of the synthesis mix with crude rabbit reticulocyte lysate in a standard assay volume of 200 l at 37°C. Each degradation assay contained 40 mM Tris/HCl, pH 7.5, 5 mM MgCl 2 , 2 mM dithiothreitol, 0.5 mM ATP, 10 mM phosphocreatine, 0.05 mg/ml creatine phosphokinase, 0.1 mM cycloheximide, and 50 l of reticulocyte lysate unless stated otherwise.
In experiments testing an energy requirement for SSAT degradation, ATP and the ATP regenerating system were replaced by 20 mM 2-deoxyglucose and 10 g/ml hexokinase. In experiments testing the effect of antizyme on the reaction, a polyhistidine-tagged preparation of antizyme was used. This was obtained by inserting the antizyme cDNA containing a deletion of the thymidine present at position 205 (34) into the pQE-30 plasmid, expressing the protein in E. coli and purifying the resulting protein to homogeneity by immobilized metal affinity chromatography.
The rate of SSAT or ODC degradation was followed by removing 30-l aliquots from the degradation assay at timed intervals as shown in the legends to the figures. Aliquots were mixed with SDS sample buffer and boiled for 10 min prior to being resolved by SDS-PAGE. The rate of 35 S-labeled protein degradation was determined by quantifying the fixed and dried gels using a Molecular Dynamics 425E-120 Phos-phorImager and ImageQuant application software.

Degradation of SSAT and ODC in an ATP-dependent Reticulocyte Lysate
System-The degradation of SSAT was studied by using a reticulocyte lysate system shown previously to degrade ODC in a physiologically relevant manner (29,33,35). Labeled SSAT was synthesized using the TNT synthesis system with pSAT9.3, and aliquots were added to reticulocyte lysates. As shown in Fig. 2, SSAT was degraded very rapidly by these lysates, and the loss of the SSAT band occurred at a rate comparable to that of ODC. A maximal rate of ODC degradation required the addition of antizyme, which is known to be present in limited amounts in reticulocyte lysates (26), whereas antizyme had no effect on the rate of loss of SSAT.
The rapid degradation of the 35 S-labeled SSAT required the presence of ATP and an ATP-regenerating system (results not shown). SSAT was stable during a 3-h incubation when ATP was omitted from the reaction and 2-deoxyglucose and hexokinase was included to deplete ATP pools present in the lysate. Substitution of ATP and the ATP-regenerating system with 2 mM of the nonhydrolyzable ATP analog, ATP␥S, also resulted in a marked stabilization of SSAT with only a 30% loss in FIG. 1. Amino acid sequence of human SSAT. Sites of mutations described in the text are indicated with bold characters. Residues that are essential for BE-3-4-3 induced stabilization are shown in bold, the glutamic acid residues found to be essential for rapid degradation are shown in bold with double underlining, and the lysine residue whose mutation was found to increase the rate of degradation is shown in bold with single underlining. The sites of cleavage of SSAT by trypsin or protease Glu-C are shown in italics.

FIG. 2. Effect of antizyme on the degradation of ODC and
SSAT. The 35 S-labeled wild type ODC (filled symbols) and wild type SSAT (open symbols) proteins were synthesized in the TNT-coupled transcription/translation system and used as substrates for degradation in the presence of an ATP regenerating system as described under "Experimental Procedures." The proteins were incubated in the degradation assay at 37°C in the presence of 0 (circles), 2 (squares), or 20 (triangles) ng of purified recombinant histidine-tagged antizyme. Aliquots of 30 l were removed from the degradation assay at the times indicated, and proteolysis was determined by quantifying the intact labeled protein remaining following separation by SDS-PAGE and quantitated as described under "Experimental Procedures." intensity of the 20-kDa band after 3 h of incubation. This small loss is likely to reflect the depletion of preexisting ATP pools in the lysate, as most of the degradation seen occurred early in the incubation.
Ubiquitin-mediated Degradation of SSAT-Incubation of the 35 S-labeled SSAT in reticulocyte lysates led to the appearance of a ladder of higher molecular weight bands of labeled protein, which were readily visible in the films developed from the SDS-PAGE analysis (Figs. 3 and 4). The characteristic ladder was noticed to be more intense at earlier times, becoming fainter with extended incubation in the degradation reaction. To confirm that the higher molecular weight bands were associated with SSAT, aliquots taken at timed intervals from the degradation reaction were mixed with a polyclonal antibody to SSAT, and the immune complexes were precipitated upon addition of protein A and resolved by SDS-PAGE (Fig. 3). The higher molecular weight bands were precipitated by this antibody showing that the ladder does represent SSAT complexes. The size of the two lowest molecular weight complexes are consistent with the predicted size of mono-and di-ubiquitinated SSAT, suggesting that SSAT may be degraded via the ubiquitin-proteasome pathway.
As shown in Fig. 4, the rate of degradation of SSAT was reduced by the addition of calpain inhibitor I, which is one of a class of peptide aldehydes reported to be inhibitors of 20 S proteasomal function (36). This inhibitor also caused a persistence in the SSAT-ubiquitin bands. Since calpain inhibitor I is not absolutely specific for proteasomal proteases, the effect of a fragment of human S5a protein was also studied. The S5a protein (37) is a component of the 26 S proteasomal complex that has been shown to recognize and bind polyubiquitinated substrates (38,39), thus targeting them for degradation by the 26 S proteasome. Rechsteiner and colleagues (40) showed that this protein acts as a potent and specific inhibitor of ubiquitindependent proteolysis. A fragment of this protein that contains the portion responsible for the inhibitory activity blocked SSAT degradation in a dose-dependent manner (Fig. 4). The characteristic ladder of ubiquitinated SSAT associated with SSAT degradation persisted longer in the presence of the S5a fragment (Fig. 4A). The addition of 1.5 mg/ml S5a fragment led to about 58% of the 20-kDa band corresponding to SSAT remaining after a 3-h incubation at 37°C (Fig. 4B).

Effect of Polyamines and BE-3-4-3 on SSAT Degradation-
The degradation of 35 S-labeled SSAT became progressively slower with increasing concentrations of spermidine and spermine added to the degradation assay (Fig. 5). Spermine was more effective than spermidine in preventing SSAT degradation with 73% compared with 36% of the 20-kDa band remaining after a 2-h incubation in the presence of 1 mM spermine or spermidine, respectively. BE-3-4-3 was more potent than either of the natural polyamines, giving maximal protection of SSAT from degradation in the reticulocyte lysate system at 100 M concentrations (Fig. 6). It is known that either BE-3-4-3 (22) or much higher albeit physiological levels of natural polyamines 2 produce a conformational change in SSAT structure, and the results shown in Figs. 6 and 7 therefore indicate that this conformational change renders SSAT resistant to degradation by the ubiquitin-dependent system present in the reticulocyte lysates. Addition of BE-3-4-3 did not prevent the ubiquitination of the SSAT protein, since the higher molecular weight bands were still present when either wild type or the K166A mutant SSAT described below were incubated with the lysates supplemented with BE-3-4-3 (re-FIG. 3. Immunoprecipitation of SSAT complexes from the degradation assay. Assays were carried out as in Fig. 2, and at the times shown, 30-l aliquots were removed and analyzed. The samples were not treated (left lanes), treated at 4°C for 1 h with a polyclonal anti-SSAT antibody (center lanes), or treated in the same way with preimmune serum (right lanes). The antibody-or control serum-treated samples were then incubated with 30 l of a 10% washed suspension of protein A for an additional 1 h at 4°C, and the immune complexes were collected after centrifugation at 13,000 rpm for 5 min. The supernatant was discarded, and the pellets were washed twice with buffer (50 mM Tris/HCl, pH 7.5, 0.1 mM EDTA, 2.5 mM dithiothreitol) prior to preparation for electrophoresis. sults not shown).
Effect of Mutations in SSAT on Proteasomal Degradation-To investigate the structural features of SSAT responsible for the rapid proteasomal degradation and for the prevention of this degradation by BE-3-4-3, a series of alterations to the SSAT sequence (Fig. 1) were made by site-directed mutagenesis, and the mutant proteins were incubated for varying periods of up to 3 h in the presence or absence of BE-3-4-3. Results for the amount of degradation that occurred in 1 h, which were representative of the results of the entire time courses, are shown in Fig. 7. The gels showing results for selected mutants of particular interest are shown in Fig. 6.
The rapid degradation of SSAT in the absence of BE-3-4-3 requires the carboxyl end of the molecule, since mutants E170Stop, A168Stop, and M167Stop were not rapidly degraded (Fig. 7A). The critical importance of the carboxyl side chain of the terminal two glutamic acid residues is shown by the results with point mutations where either was altered to glutamine. Both of these mutants (E170Q and E171Q) were stable (Figs. 6A and 7A). Replacing these two acidic residues with basic lysines (mutant E170K/E171K) or adding two lysines as additional residues to the end of the SSAT sequence (mutant ϩ172K/173K) also rendered the SSAT stable (Fig. 7A). The SSAT mutants rendered stable by mutation or truncation of the carboxyl domain were still subject to ubiquitination since the higher molecular weight bands similar to those seen in Figs. 3 and 4 were still seen with mutants E170Stop, E171Q, E170Q, and E170K/E171K (results not shown). Only changes in the carboxyl end of the protein caused the SSAT to become resistant to proteasomal degradation. None of the other point mutations studied (which encompass residues throughout the molecule) prevented the rapid degradation of the SSAT protein (Fig. 7). The triple mutant K141S/R142A/R143S did prevent rapid degradation of SSAT (Fig. 6B). This mutation removes the site of tryptic digestion (22) of pure SSAT but may distort the structure of the protein since this mutation also renders the protein completely inactive.
Mutant K166A was degraded even more rapidly than the wild type protein. This effect is seen in the data shown in Fig.  7A but is underestimated because even the control protein is substantially degraded in the 1-h time period. Fig. 6A, in which the entire time course is shown, indicates clearly that this mutation substantially increases the rate of degradation of the SSAT.
However, this K166A mutation did not prevent the ability of BE-3-4-3 to stabilize the SSAT protein. Several mutations did reduce or totally abolish the protective effect of BE-3-4-3. These included mutants M167A, R155A, E152K (and E152Q), H126A, C122A, and R19A (Figs. 6B and 7). All of these mutations have been shown to reduce the ability of SSAT to bind polyamines or polyamine analogs (20,22). Mutations R7A, C14A, and K141A also prevented the stabilization by BE-3-4-3 (Fig. 7, B and C). In addition to providing information on the residues responsible for stabilization by BE-3-4-3, these observations rule out the possibility that BE-3-4-3 inhibits the degradation system directly.

DISCUSSION
Evaluation of the properties of protein mutants generated by site directed mutagenesis is sometimes subject to question because of the possibility that the protein structure is distorted as a result of the mutation. However, this is unlikely to account for our results since the key mutations have previously been tested and shown to produce little or no alteration in SSAT enzymatic activity or in the characteristic pattern of SSAT sensitivity to proteases (20,22). Previous studies have shown that SSAT shows a distinctive and limited degradation when exposed to trypsin or to protease Glu-C and is cleaved only at discrete sites which are -Lys 141 Arg 142 Arg 143 -and -Glu 151 -Glu 152 -, respectively (Fig. 1). Degradation by these proteases is completely blocked by addition of BE-3-4-3 (22). Mutations E170Q and E171Q, which are shown in the current experiments to abolish the rapid degradation, and mutation K166A, which increases the rate of proteasomal degradation, produced little or no reduction in SSAT enzymatic activity and no change in the binding of BE-3-4-3 or polyamines (measured by the ability to protect SSAT from proteolytic degradation or by the K m for spermidine) (22).
The results reported here were obtained with a reticulocyte lysate system in vitro and are entirely consistent with the known facts about SSAT degradation in vivo. The half-life of SSAT in polyamine-depleted and control cells was estimated at 9 and 20 -40 min, respectively (15)(16)(17) and was increased to Ͼ12 h by the presence of polyamine analogs (15,18). The findings that SSAT degradation by the ubiquitin-proteasomal system is greatly reduced by BE-3-4-3 is in agreement with these observations. It appears likely that this protection is mediated by an alteration in configuration in the SSAT protein brought about by binding the polyamine analog. Most of the mutations that were found in the experiments illustrated by Figs. 6 and 7 to abolish the stabilization of SSAT by BE-3-4-3 have also been shown to abolish the protection of the SSAT protein from protease cleavage and to increase the K m for spermidine in the acetylation reaction. These include the mutants M167A, R155A, E152K or E152Q, H126A, C122A, and R19A (20,22). Therefore, it is likely that the inability of BE-3-4-3 to stabilize these mutant SSATs in the proteasomal degradation system is due to a lack of binding of the polyamine analog. Although these mutations do have reduced levels of SSAT activity, it is unlikely that they cause major alterations in the structure of SSAT since (a) the major part of the reduction in activity is due to a change in the K m for polyamines, and (b) the mutant E152K forms enzymatically active heterodimers when co-expressed with mutant R101A which forms part of the acetyl-CoA binding site (20).
A simple model that would account for our findings would be that SSAT degradation is brought about by ubiquitination of the protein that causes it to be bound to the proteasomal structure. Degradation then requires the interaction of the proteasome with the carboxyl end of the SSAT, and this interaction requires the carboxyl side chains of the glutamic acid residues 170 and 171. The binding of BE-3-4-3 or polyamines alters the configuration of the protein so that this carboxyl end is not exposed and the degradation is therefore prevented. Several peptide motif sequences have been proposed to target intracellular proteins for rapid destruction (41). The terminal -MATEE motif of the SSAT appears to provide another example of such a sequence. The most widespread motif that has been hypothesized to mark proteins for rapid turnover is the PEST sequence, which is defined as a region of Ն12 amino acids that contains proline, serine/threonine, and glutamic acid/aspartic acid in the absence of basic residues (41,42). The PEST hypothesis has received much support from the almost ubiquitous presence of such PEST sequences in rapidly turning over proteins, although it has not yet been determined how the sequences act. SSAT (Fig. 1) does not contain an obvious PEST sequence. The terminal -MATEE motif, which is identified in the current study as a critical region for SSAT turnover, is eliminated from consideration by the absence of a proline residue, but the other components of the PEST motif, acidic residues and serine or threonine, are present in the carboxylterminal sequence, and the carboxyl-terminal location may permit this sequence to be exposed for interaction with the proteolytic machinery without the need for a proline. Furthermore, positively charged residues are not compatible with PEST sequences, and the mutation to alanine of lysine 166, which increases the liability of SSAT, extends the length of the acidic carboxyl-terminal sequence. It is therefore possible that -MATEE acts as a "pseudo-PEST" sequence.
Although both enzymes lead to an increase in putrescine in the cell, ODC and SSAT have opposing roles in polyamine synthesis, with ODC tending to increase and SSAT to decrease polyamine levels (24). Both enzymes are regulated at the level of protein stability by polyamines but in opposite directions. ODC content is reduced by polyamines via the antizyme-mediated enhancement of degradation, whereas SSAT is increased by polyamines via the prevention of degradation. In both cases, the carboxyl end of the protein is a critical region for interaction with the proteasome. Removal of from 5 to 37 residues from the carboxyl terminus of ODC (43,44), or the point mutation of cysteine 441, which is contained within this region FIG. 7. Effect of mutations on SSAT degradation and protection by BE-3-4-3. The 35 S-labeled wild type and mutants of SSAT were incubated in the reticulocyte lysate degradation assay in the absence and presence of 100 M BE-3-4-3 as in Fig. 6. The results are shown as the percent of SSAT protein degraded in 1 h. The mutations used are shown and the results are plotted in order of mutations from the carboxyl to the amino terminus of the protein. Panel A shows results for the region covering amino acids 166 -171, panel B shows results for the region covering amino acids 122-161, and panel C shows results for the region covering amino acids 7-120. (45), renders ODC stable without affecting its enzymatic activity. Attaching the same 37 residues of ODC to the terminus of dihydrofolate reductase rendered this protein subject to rapid degradation (46). It will be of interest to determine whether the carboxyl-terminal domain of SSAT is also able to impart rapid turnover to other proteins.