INTRODUCTION

Spermidine/spermine N¹-acetyltransferase (SSAT)¹ is the rate-limiting step in polyamine catabolism that catalyzes the transfer of the acetyl group from acetyl-CoA to the N¹ position of spermidine or spermine (1, 2, 3). The acetylated polyamines are then excreted from the cell in their acetylated forms or are metabolized by the constitutive FAD-requiring polyamine oxidase. SSAT is a key component in the homeostatic system that maintains a narrow range of intracellular polyamines in eukaryotic cells. SSAT is normally induced in response to conditions that lead to rapid increases in the natural polyamines and to stress stimuli (4, 5, 6, 7). The superinduction of SSAT by a series of antitumor polyamine analogues has been implicated in the cell type-specific cytotoxic response of several important human solid tumors (8, 9, 10, 11). Since several of the polyamine analogues are now in or are being considered for clinical trial, a greater understanding of the role of SSAT and its precise mechanism of action is critical for the better design and use of these new antineoplastic agents.

Although a kinetic model for the activity of this enzyme has previously been postulated by Della Ragione and Pegg (12) to be an ordered Bi Bi molecular reaction where the natural polyamine is bound first and the acetylated polyamine is released last, little is known about the specific molecular domains required for activity. The molecular cloning of the human SSAT gene, cDNA, and microsequencing of most of the purified human protein have allowed a more detailed study of the residues and domains required for enzyme activity (13, 14, 15, 16). The human SSAT gene codes for a 171-amino acid protein with a predicted molecular weight of ~20,000 and is thought to be active as a homotrimeric or homotetrameric protein (17). SSAT protein and enzyme activity are normally undetectable in cells that have not been induced. However, in polyamine analogue-responsive cells, the superinduction of SSAT expression can result in a >1000-fold increase in protein activity that is entirely a result of new protein synthesis (8). In extreme cases, the induction can result in SSAT approaching 1% total cellular protein (13). The induction of SSAT is a result of a complex series of events that starts with the induction of transcription of the SSAT gene and involves the stabilization of the SSAT mRNA, increased translational efficiency, and stabilization of the SSAT protein by the natural polyamines and their analogues (5, 14, 16, 18, 19, 20). Coleman et al. (20) have recently reported on a domain of the human SSAT protein that appears to be responsible for the stabilization of SSAT by the polyamine analogue N¹,N¹²-bis(ethyl)spermine and may additionally act as part of the natural polyamine-binding/catalytic site.

In this study, we attempted to determine what regions of the human protein are required for acetyl-CoA binding and enzyme activity of the human SSAT protein. Using a combination of site-directed mutagenesis, in vitro transcription/translation, and extensive data base homology searching, we identified a region of high homology in the human SSAT protein that is necessary for activity and is highly conserved among several microbial antibiotic N-acetyltransferases that use acetyl-CoA as a substrate. The data in the current study demonstrate that the human SSAT protein can be considered as a member of an evolutionarily conserved superfamily of N-acetyltransferases.

EXPERIMENTAL PROCEDURES

The human SSAT cDNA (14) was used for all in vitro mutagenesis. [-³⁵S]dATP (1000 Ci/mmol) and [³⁵S]methionine were purchased from Amersham Life Science, Inc. For SSAT assay to determine the activity of mutant SSAT, [1-¹⁴C]acetyl-CoA (58 mCi/mmol) from Amersham was used. Restriction and DNA-modifying enzymes were purchased from Life Technologies, Inc., and New England Biolabs, Inc. (Beverly, MA). Altered Sites® II in vitro mutagenesis systems and TNT® coupled transcription/translation reticulocyte lysate systems were purchased from Promega (Madison, WI). Other chemicals were purchased from Sigma, Boehringer Mannheim, and J. T. Baker Inc.

A fragment of the pSATH2 cDNA containing the entire coding region of the SSAT protein (14) was produced by restriction with KpnI and XbaI. The fragment was then inserted into the pAlter-1 vector (Promega) in the T7 promoter orientation. The ligation product was introduced into JM109 Escherichia coli strain by electroporation with an E. coli pulser (Bio-Rad), according to the instructions of the manufacturer.

Site-directed mutagenesis was performed using the Altered Sites II mutagenesis system from Promega, as previously reported (21) with the mutagenic primers listed in Table I. The mutagenized plasmid was then introduced into the repair-deficient strain of E. coli ES1301 mutS for replication, expansion, and plasmid preparation. Wizard^TM minipreps (Promega) of the mutant plasmids were then introduced by electroporation into the JM109 strain of E. coli, and the sequence of each mutant was confirmed by the dideoxy chain termination method of sequencing (22) using the Sequenase version 2.0 sequencing kit (Amersham Life Science, Inc.).

Table I. Mutant nucleotides used for site-directed mutagenesis Desired mutation Mutant oligonucleotide sequencesa H49A 5-GGAGACCTTTTACCAC-3 H53A 5-CCTTTTACCTGCCTGGTTGC-3 H63A 5-CGAAAGAGCTGGACTCCGG-3 H69A 5-GAAGGACAGCATTGTTGG-3 E92A 5-GTATCTTGGGACTTCTTCTTCG-3 D93A 5-CTTGAGGCTTCTTCGTGATC-3 F94A 5-CTTGAGGACCTTCGTGATGAG-3 V96D 5-GGACTTCTTCGACATGAGTG-3 D99A 5-GATGAGTGTTATAGAGGCTTTG-3 R101A 5-GAGTGATTATAGGCTTTGGC-3 G102D 5-GATTATAGAGCTTTGGCATAG-3 G104D 5-GAGGCTTTGCATAGGATCAG-3 I105A 5-GAGGCTTTGGCAGGATCAGAAATTC-3 G106D 5-CTTTGGCATAGTCAGAAATTC-3 G144D 5-CTATAAAAGAAGAGTGCTTCTG-3 a  The bases representing the nucleotide mutations are underlined.

Plasmids containing mutant SSAT cDNA were purified by the Qiagen Maxi Prep method (Qiagen, Inc., Chatsworth, CA) according to the instructions of the manufacturer. Purified wild-type and mutant plasmids were used to produce protein in the TNT coupled transcription/translation lysate system from Promega according to the manufacturer's supplied protocol. Relative amounts of translated products for comparison of wild-type and mutant SSAT were determined by PhosphorImager analysis after performing parallel translations in the presence of [³⁵S]methionine followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (23) on a Molecular Dynamics PhosphorImager using Image Quant software (Sunnyvale, CA). The activity (described below) of each assay was normalized by comparison with the amount of labeled protein corresponding to the parallel translation for each assay.

Relative activity and acetyl-CoA K_m determinations were performed essentially as previously reported (8, 13). Appropriate dilutions were made in 5 mM Hepes buffer (pH 7.2) and 1 mM dithiothreitol. The standard assay system contained 3 mM spermidine and 8.8 µM acetyl-CoA. For K_m determinations of acetyl-CoA, the spermidine concentration was fixed at 3 mM and the acetyl-CoA was varied between 2 and 10 µM as indicated under ``Results.''

The blastp program available through the BLAST E-mail server (blast{at}ncbi.nlm.nih.gov) and the PredictProtein program (PredictProtein{at}EMBL-heidelberg.de) were used for protein homology and secondary structure predictions, respectively. The sequences used for comparison were: SAT4_CAMCO, Campylobacter coli streptothricin acetyltransferase; STAT_ESCO, E. coli streptothricin acetyltransferase; STAT_STRLA, Streptomyces lavendulae streptothricin acetyltransferase; ACYT_BACSU, Bacillus subtilis acyltransferase; NACT_STRNO, Streptomyces noursei nourseothricin acetyltransferase; GNAT_SACCE, Saccharomyces cerevisiae gentamicin 3-acetyltransferase; STA_TPTN7, transposon Tn7 streptothricin acetyltransferase; GPACT_SACCE, L-A virus GAG protein N-acetyltransferase; YSCCHRV_52, S. cerevisiae hypothetical 18.1-kDa protein in SNP2-MDJ1 intergenic region; GNAT_PSEAE, Pseudomonas aeruginosa gentamicin acetyltransferase; AMAT_PSEAE, P. aeruginosa aminoglycoside 3-N-acetyltransferase; AACA_SHISO, transposon Tn2426 (from Shigella sonnei) 6-N-acetyltransferase; AMAT_ACIHA, Acinetobacter haemolyticus aminoglycoside 6-N-acetyltransferase; AMAT_ENTAE, Enterobacter haemolyticus aminoglycoside 6-N-acetyltransferase; SPAT_ESCCO, E. coli spermidine acetyltransferase; SNAT_RAT12, rat pineal serotonin N-acetyltransferase; ANAT_OVINE, ovine pineal serotonin N-acetyltransferase.

RESULTS

At the initiation of this project, little other than the primary amino acid structures of the human, mouse, and hamster SSATs were known. Although there is a high degree of conservation among the mammalian SSAT enzymes, no significant homology had been observed between these and other known proteins, including the recently cloned and characterized E. coli spermidine acetyltransferase (14, 24, 25, 26). Therefore, to determine whether any smaller regions of homology exist between the human SSAT and other acetyltransferases, we used the blastp program available through the BLAST E-mail server. Narrowing the query parameters to a 20-amino acid region corresponding to the human SSAT amino acids Tyr-90-Leu-110, the blastp program identified 14 acetyltransferases of a microbial antibiotic acetyltransferase superfamily (27) with high homology to the human SSAT. The highest homology in this region, defined as Domain I (Table II), was found between the C. coli streptothricin acetyltransferase where there are 10 identical amino acids and 6 highly conserved residues. Each of the acetyltransferases found to be homologous to SSAT is N-acetyltransferases as opposed to O-acetyltransferases. The most significant area of homology occurs surrounding the RGFGIGS region of Domain I beginning at Arg-101 of SSAT. Closer inspection of the amino acids surrounding this region led to the identification of a second domain of lesser homology (Table II) ranging from Asn-133 to Gly-144 in the human SSAT gene. Domains I and II are typically separated by 22 amino acids in the human SSAT as is the case for most of the antibiotic acetyltransferases. The recently cloned E. coli spermidine N-acetyltransferase (27) had less homology to the human SSAT than most of the antibiotic acetyltransferases. During the preparation of this article, Borjigin et al. (28) and Coon et al. (29) reported the cloning of rat and sheep N-serotonin acetyltransferase, respectively. Each of these mammalian acetyltransferases contains both Domain I and Domain II as was recognized by Coon et al. (29). These sequences have been included for comparison with SSAT and the antibiotic acetyltransferases.

Table II. Amino acid homologies between human spermidine/spermine N1-acetyltransferase and other N-acetyltransferases Sequence name SWISSPROT (s) or GenBank (g) accession no. Starta Domain Ib Spacec Domain II Endd Consensuse UEDUUVXXXURGXGUG$XUU 22 NXPAUXUYXRUG SSAT_HOMSA M55580g 90 LEDFFVMSDYRGFGIGSEIL 22 NEPSINFYKRRG 27 SAT4_CAMCO UO1945g 98 IEDIAVCKDFRGQGIGSALI 22 NLIACKFYHNCG 28 STAT_ESCCO Z48231g 97 IDDIAVDVPYRGSGVSRLLM 22 NLAACRFYRRYG 29 STAT_STRLA M16813s 104 IEDIEVAPGHRGKGIGRVLM 22 NAPAIHAYRRMG 29 ACYT_BACSU P39909s 78 LDRFFIIERYQGKGLGKKML 23 NIHAIRLYQRFG 19 NACT_STRNO S60706g 107 VEDIEVAPEHRGHGVGRALM 22 NAPAIHAYRRMG 30 GNAT_SACCE U25841g 89 INDLYVDENSRVKGAGGKLI 22 NHRAQLLYVKVG 13 STAT_TPTN7 M63169g 85 IEHIVVSHTHRGKGVAHSLI 22 NVPACNLYAKCG 35 GPAT_SACCE Q03503g 80 IGMLAVESTYRGHGIAKKLV 22 NSAALNLYEGMG 42 YSCCHRV_52 P43577s 95 IEDIAVNSKYQGQGLGKLLI 19 DEKNVKFYEKCG 12 GNAT_PSEAE U12338s 107 IYDLAVSGEHRRQGIATALI 22 DDPAVALYTKLG 16 AMAT_PSEAE L06157g 106 IYDLAVASSHRRLGVATALI 22 DDPAVALYTKLG 16 AACA_SHISO M86913g 79 LHPLVVRPDYQNKGIGKILL 48 NKHPYEFYOKNG 27 AMAT_ACIHA L09246g 76 LEGIYVLPAHRRSGVATMLI 22 NVISHAMHRSLG 15 AMAT_ENTAE U13880g 84 LEGWYVVPSARRQGVGVALV 22 NSASTSAHLAAG 14 SPAT_ESCCO D25276g 84 EFQIIISPEYQGKGLATRAA 21 NEKAIHIYRKLG 47 SNAT_RAT12 U40803g 118 LHVLAVHRTFRQQGKGSVLL 20 ENALVPFYEKFG 35 ANAT_OVINE U29663g 120 LHALAVDRSFRQQGKGSVLL 20 EDALVPFYQRFG 35 Secondary Structuref .....EEE....LLLLHHHH ......HHH.H. a  Number of amino acids preceding Domain I. b  Amino acids which are identical to or conserved with the human SSAT sequence are in bold type. c  Number of amino acids between Domain I and Domain II. d  Number of amino acids to the carboxyl terminus of each protein. All listed acetyltransferases are of similar size. e  Consensus amino acid sequence of the N-acetyltransferases. Single letter designation is used when appropriate. U, bulky hydrophobic; S, serine or threonine; X, any amino acid. f  Secondary structure predicted for human SSAT protein. E, -sheet; L, glycine loop; H, -helix.

Predicted Secondary Structure Suggests That a -Sheet-Loop--Helix Motif Is Required by Human SSAT for Acetyl-CoA Binding

The PredictProtein program defined a consistently shared feature (14 of 14 compared sequences) between SSAT and the antibiotic acetyltransferases. An -helix structure begins at the RGFGIGS region of SSAT and near the homologous regions in each of the other acetyltransferases. In the human protein, Domain I is characterized by a -sheet-glycine loop--helix (Table II). A similar secondary structure is predicted for the E. coli spermidine N¹-acetyltransferase even though the amino acid homology in E. coli protein is less than that observed in many of the antibiotic acetyltransferases (data not shown). The initiation of an -helix is not predicted for the sheep serotonin N-acetyltransferase (ANAT_OVINE). In contrast to Domain I, the predicted secondary structure for Domain II is not conserved among the various acetyltransferases. Based on confirmed secondary structure analysis, the PredictProtein model has been demonstrated to be ~71% accurate (30).

Since the greatest homology between the human SSAT protein and the acetyltransferases occurs in Domain I, this area was chosen for the majority of mutations in this study. The most conserved residues among the proteins examined corresponded to Val-96, Arg-101, Gly-104, and Gly-106 of the human SSAT protein. Mutations to any of these amino acids produced a protein with no measurable activity (Table III). Gly-102, also highly conserved among these acetyltransferases, when changed to an aspartate resulted in a protein with only 23% SSAT activity compared to the wild-type protein. The elimination of activity was not due to differences in the amount of translated proteins analyzed from these mutations, since protein amounts were normalized by protein amount as determined by PhosphorImager analysis of [³⁵S]methionine-labeled protein (Fig. 1). In addition, the E92A and D93A mutations significantly reduced the activity of the resultant proteins, suggesting that these negatively charged residues may be necessary for activity. Only the V96A mutation produced a significant change in the predicted secondary structure of the human SSAT protein. This mutation resulted in the complete loss of the most proximal -sheet and a significant reduction of the glycine loop (Table II). The other mutations of highly conserved residues produce only modest changes in the predicted secondary structure.

Table III. Analysis of in vitro transcription/translation SSAT mutant proteins SSATa Relative activityb Acetyl-CoA (Kmc) % µM Wild-type 100 3.0 H49A 55  ± 5.8 5.4 H53A 105  ± 2 5.8 H63A 81  ± 8 6.0 H69A 117  ± 2 10.0 E92A 5.5  ± 0.2 3.3 D93A 12.9  ± 0.2 NDd F94D <1 ND V96D <1 ND D99A 30  ± 0.7 5.0 R101A <1 ND G102D 30  ± 3.7 35.7 G104D <1 ND I105A 67  ± 7.0 6.1 G106D <1 ND G144D 4.5  ± 6.01 41.9 a  Indicates SSAT mutant analyzed. b  Results of a representative experiment with four replicates comparing SSAT mutants with the wild-type protein. The spermidine concentration was fixed at 3 mM and the acetyl-CoA concentration was fixed at 8.8 µM. Values are mean ± S.D. c  Km determinations for mutant and wild-type SSAT. Values represent Kn determined by Lineweaver reciprocal plot with a 3 mM fixed concentration of spermidine and varying the acetyl-CoA concentration from 2 to 10 µM. r2 for the linear regression in each Km determinations was >0.95. d  ND, not determined.

The most conserved residue in Domain II is represented by Gly-144. This residue is found in all of the microbial acetyltransferases. However, in the hamster SSAT, the Gly-144 is replaced by alanine (24). The G144D mutation reduced SSAT activity by >95% (Table III).

Since most of the mutations to the highly conserved residues in Domain I produced inactive proteins, the studies were limited to the proteins that maintained measurable enzymatic activity. Although the in vitro transcription/translation products were used for the kinetic analysis presented here and the in vitro translation products may contain impurities which can have an effect on the kinetic analysis, the K_m values determined for the wild-type protein produced in the TNT are comparable with those previously obtained using purified human SSAT protein (13, 17).

The apparent K_m for acetyl-CoA of each of the partially active mutants was determined (Table III). To determine the K_m of acetyl-CoA, the spermidine concentration (3 mM) was fixed well above the concentration that saturates the wild-type protein, and the concentration of acetyl-CoA was varied from 2.0 to 10 µM. Mutation of each of the several histidine residues of the human SSAT protein resulted in proteins that maintained significant activity, and of these, only H69A demonstrated a significant change in K_m for acetyl-CoA (Table III). The K_m for acetyl-CoA was significantly affected by mutation of two of the most conserved glycines, G102D and G144D. These mutations each resulted in proteins with an apparent K_m for acetyl-CoA >10-fold higher than the wild-type translation product (35.7 and 41.9 µM, respectively, compared to 3.0 µM measured for the wild-type protein). These results are consistent with the RGFGIGS region in Domain I and Gly-144 of Domain II in SSAT being an important determinant of acetyl-CoA binding. Preliminary studies indicate that the K_m for spermidine is also significantly increased for the G144D mutant as compared to the wild-type protein (data not shown). It is possible that the mutation of G144D results in a general disruption of enzyme structure, thus leading to the observed decrease in affinity for both acetyl-CoA and spermidine.

DISCUSSION

SSAT is the rate-limiting enzyme in polyamine catabolism, and its superinduction appears to be associated with sensitivity to certain antitumor polyamine analogues (1). Its recent cloning has provided a means to examine the specific regions of the protein required for enzyme activity and to gain a better understanding of potential structural targets for use in antitumor drug development.

Our goal was to identify those regions required for the acetyl-CoA binding. The amino acids chosen for site-directed mutagenesis studies were based on multiple criteria. Histidine residues have been implicated in the activity of other acetyltransferases, particularly choline acetyltransferase (31, 32). Most importantly, sequence analysis of the human SSAT protein revealed a significant homology with highly conserved domains of a superfamily of microbial antibiotic acetyltransferases (33).

Our current work clearly defines a region of homology between distantly related N-acetyltransferases, referred to here as Domain I, that contains the RGFGIGS motif and is critical for the activity of the human acetyltransferase. Although similar motifs have been postulated to be important in acetyl-CoA binding (33), there are no reports that have provided site-directed mutational data to support this hypothesis. Della Ragione et al. (34) predicted that the acetyl-CoA binding site would contain at least one arginine. This hypothesis was based on treatment of purified rat SSAT with reagents known to react with arginine residues. An arginine residue has also recently been demonstrated to be critical in the binding of acetyl-CoA to the active site of rat choline acetyltransferase (35). The results demonstrating that the R101A mutant is completely inactive supports the hypothesis that an arginine is part of the active site of human SSAT. It is also not surprising that the human SSAT protein bears greatest homology to the antibiotic acetyltransferase with a targeted nitrogen that is associated with a moiety most like the polyamines in structure.

Glycine loops have been postulated to be critical in the binding of mono- and dinucleotide substrates, and Schulz (27) has proposed that lack of bulky side chains typified by glycine loops are necessary to allow the nucleotide binding without steric hindrance. A common feature of the N-acetyltransferases is the -helix that begins in the glycine loop area homologous to the RGFGIGS motif of SSAT. Both the relative activity and preliminary kinetics from the mutants strongly suggest that this region is the site of acetyl-CoA binding. However, final confirmation of the importance of this region is that SSAT activity will require crystallographic analysis of the purified human SSAT.

Domain II appears to be less conserved among the various representative acetyltransferases. However, two residues are highly conserved in this region. The glycine corresponding to Gly-144 is conserved among all of the microbial acetyltransferases and the mammalian serotonin N-acetyltransferases. The hamster SSAT (24), which only has six differences in amino acid structure and no differences between the human SSAT in Domain I, has an alanine in the 144 position. Tyr-140 is the second most conserved residue in Domain II and suggests the possibility that tyrosine phosphorylation may have some role in the activity or regulation of the acetyltransferases. During the preparation of this article, Borjigin et al. and Coon et al. reported the cloning and primary structure of rat (28) and sheep (29) serotonin N-acetyltransferases. Each of the proteins conforms to the consensus regions reported here, and Coon et al. (29) presented similar domains which they defined as Motifs I and II.

Recent work by Coleman et al. (20) convincingly demonstrated that the most carboxyl-terminal region of SSAT, consisting of the amino acids MATEE, is critical for polyamine and polyamine analogue binding which in turn lead to stabilization of SSAT. Their results also suggest that this region may be critical for catalytic activity. Mutations of acidic residues different from those reported by Coleman et al. (20), E92A and D93A, produced proteins with activities reduced >85%. These acidic residues may be similar in function to the glutamates in the MATEE region with regard to polyamine binding and enzyme activity.

In the current study, the effects of mutations of the putative acetyl-CoA binding site on enzyme stability were not studied. It is unlikely that the stability of SSAT would be controlled by intracellular acetyl-CoA concentration since polyamine homeostasis, not acetyl-CoA regulation, is the primary function of SSAT. Although SSAT protein is routinely prepared in the presence of high polyamine concentrations for maintenance of stability, acetyl-CoA is not usually added, suggesting that it is not required for enzyme stability. However, some of the mutations made in the current study may significantly affect enzyme conformation, which could lead to a less stable protein.

In summary, the mutation of several residues in Domain I, which is presented as a putative acetyl-CoA-binding site of the human SSAT, leads to inactivation of the enzyme. The increased K_m for acetyl-CoA in the partially active G102D mutant strengthens the hypothesis that the RGFGIGS region is critical for acetyl-CoA binding and enzyme activity. Additionally, the lack of measurable activity of the R101A mutant is consistent with the hypothesis that an arginine is required for acetyl-CoA binding and enzyme activity (34, 35). More conservative mutations in this region along with analysis of the purified mutants will aid in further elucidation of the importance of this region in SSAT activity.