Enzyme-Substrate Intermediate at a Specific Lysine Residue Is Required for Deoxyhypusine Synthesis

Deoxyhypusine synthase catalyzes the first step in the post-translational synthesis of hypusine [Nε-(4-amino-2-hydroxybutyl)lysine] in eukaryotic translation initiation factor 5A. We recently reported biochemical evidence for a covalent enzyme-substrate intermediate involving a specific lysine residue (Lys329) in human deoxyhypusine synthase (Wolff, E. C., Folk, J. E., and Park, M. H. (1997) J. Biol. Chem. 272, 15865–15871). In an effort to evaluate the role of this enzyme-substrate intermediate in catalysis, we carried out site-directed mutagenesis (Lys to Arg and/or Ala) of the conserved lysine residues in human deoxyhypusine synthase. A drastic reduction in enzyme intermediate formation and enzymatic activities was observed with mutant proteins with substitution at Lys287 but not with those with mutations at residues 141, 156, 205, 212, 226, 251, or 338. Lys to Ala or Lys to Arg substitution at Lys329 totally abolished covalent enzyme-substrate intermediate formation and deoxyhypusine synthesis activity, indicating that Lys329 is the unique site for the enzyme intermediate and that it is absolutely required for deoxyhypusine synthesis in the eukaryotic translation initiation factor 5A precursor. The K329A mutant showed spermidine cleavage activity (∼6% of the wild type enzyme) suggesting that in contrast to deoxyhypusine synthesis, spermidine cleavage can occur without enzyme intermediate formation.

Hypusine [N ⑀ -(4-amino-2-hydroxybutyl)lysine] is formed by a unique two-step post-translational modification that occurs only in one known cellular protein, the precursor of eukaryotic translation initiation factor 5A (eIF-5A) 1 (for reviews see Refs. 2 and 3). In the first step, deoxyhypusine synthase catalyzes the NAD-dependent transfer of the 4-aminobutyl moiety from spermidine to the ⑀-amino group of a specific lysine residue (Lys 50 in the human eIF-5A precursor) of the protein substrate to form an intermediate, deoxyhypusine. Subsequent hydroxylation of this intermediate by deoxyhypusine hydroxylase completes hypusine synthesis and eIF-5A maturation (see Refs. 2 and 3).
Growing evidence supports the notion that eIF-5A and its hypusine modification are essential for eukaryotic cell proliferation (2,3). In the yeast Saccharomyces cerevisiae inactivation of eIF-5A genes (4,5) or of the deoxyhypusine synthase gene (6) 2 results in the loss of cell viability. In addition, inhibitors of deoxyhypusine synthase (7) and deoxyhypusine hydroxylase (8) exert strong antiproliferative effects in mammalian cells. The arrest of cell proliferation upon depletion of spermidine by inhibition of polyamine biosynthesis has been attributed to the consequent deficiency of hypusine and eIF-5A (9,10).
Deoxyhypusine synthases from several species share similar physical and catalytic properties. The amino acid sequence of deoxyhypusine synthase is highly conserved among the three species, human, yeast, and Neurospora crassa, for which the gene or cDNA for the enzyme has been cloned (11)(12)(13)(14)(15)(16). The native enzymes consist of tetramers of four identical subunits of 40 -43 kDa, depending on the species (11, 12, 16 -18). The enzyme catalyzes a multistep reaction leading to the synthesis of deoxyhypusine by conjugation of the 4-aminobutyl moiety of spermidine to the protein substrate (eIF-5A precursor) in a complete in vitro reaction mixture containing spermidine, NAD, and eIF-5A precursor (see Scheme 1, bold arrows). In the absence of the eIF-5A precursor, however, the enzyme catalyzes only the cleavage of spermidine to generate 1,3-diaminopropane and ⌬ 1 -pyrroline (see Scheme 1, thin arrows) (19). Both pathways are initiated by NAD-dependent dehydrogenation of spermidine to generate a postulated dehydrospermidine intermediate.
Recently we have obtained evidence for a covalent enzymesubstrate intermediate with the 4-aminobutyl moiety from spermidine attached to the ⑀-amino group of a specific lysine residue of the enzyme in an imine linkage (20 moiety, the eIF-5A precursor protein. Concurrently with the biochemical studies of the enzymeimine intermediate, we initiated site-directed mutagenesis to determine whether only a single lysine residue participates in its formation and to assess the role of the enzyme-imine intermediate in the deoxyhypusine synthase reaction. To this end we introduced a Lys to Arg and/or a Lys to Ala substitution at nine conserved lysine residues and characterized the mutant enzymes for their ability to form a tetramer and to form the enzyme intermediate and for their activity in spermidine cleavage and deoxyhypusine synthesis. The results demonstrate that a catalytically competent enzyme intermediate is formed only at Lys 329 and that this intermediate is critical for deoxyhypusine synthesis in the eIF-5A precursor protein. Our findings also suggest that, in contrast, cleavage of spermidine by deoxyhypusine synthase can occur independently of the covalent enzyme-imine intermediate. [1, H]Spermidine⅐HCl (15-27.6 Ci/mmol) was purchased from NEN Life Science Products. NaBH 3 CN from Aldrich was recrystallized (20). Oligonucleotide primers were synthesized by the Biosynthesis Company (Lewisville, TX). The pET-11a expression vector and the host Escherichia coli B strain BL 21(DE3) competent cells were from Novagen; calf intestinal alkaline phosphatase was from Promega; T4 DNA ligase was from Life Technologies, Inc.; Vent thermopolymerase was from New England Biolabs; restriction enzymes and Taq polymerase were from Boehringer Mannheim; precast polyacrylamide gels and wide range protein standards (Mark 12) were from Novex; MonoQ column and Q-Sepharose (Fast flow) resin were from Pharmacia Biotech Inc. Human eIF-5A precursor protein expressed in E. coli (ec-eIF-5A) was purified from E. coli lysates after overexpression of the human eIF-5A cDNA, as described previously (21).

Methods
Site-directed Mutagenesis-Altered enzymes in which an Arg or an Ala residue was substituted for the specified lysine residue were produced by PCR-directed mutagenesis (22) using primers in which the Lys codon (AAG) was replaced by an Arg codon (AGG for Lys 3 Arg mutations with the exception of those at Lys 329 and Lys 287 , where CGC was used) or an Ala codon (GCT) at Lys 329 or Lys 287 (Table I). Fulllength PCR products were obtained by two rounds of PCR reactions. Terminal primers A and B were designed to hybridize with the Nterminal region and the C-terminal region, respectively, with the introduction of restriction sites (NdeI in primer A and BamHI in primer B) to facilitate ligation and cloning. For each mutation a set of internal primers were designed to hybridize with the regions flanking the mutation site. Internal reverse primers (B series) have a mutated codon sequence, and half of the sequence of this primer is complementary to the forward internal primer (A series). In the first reactions, either primer A and internal primer B or primer B and internal primer A were used as a primer set, and a recombinant BlueScript plasmid containing human deoxyhypusine synthase cDNA sequence was used as a template. In the second round, primer A and primer B were used as a primer set, and two PCR products from the first round reactions were used as templates to produce a single annealed PCR product. The conditions for PCR were: 94°C for 5 min, denaturation at 94°C for 1 min, annealing at 55°C for 1 min, and extension reaction at 74°C for 2 min for 35 cycles and a final extension reaction at 74°C for 6 min. The final PCR product was digested with NdeI and BamHI and ligated to the pET11-a vector linearized with NdeI and BamHI and dephosphorylated by calf intestinal alkaline phosphatase. This ligation mixture was used to transform E. coli DH5␣. After confirming the mutation sites by sequencing, the recombinant plasmid isolated from the DH5␣ transformant was used for transformation of E. coli BL21(DE3) for protein expression.
Overexpression and Partial Purification of the Mutated Proteins-The selected transformants overexpressing the desired mutant protein were grown in 500 ml of LB medium supplemented with 100 g/ml ampicillin. When the cell density reached an optical density of 0.6 at 600 nm, isopropyl-␤-D-thiogalactoside was added to a final concentration of 1 mM, and the cells were harvested after culture for 2-3 h. Cell pellets were suspended in 10 ml of buffer A (50 mM Tris-Cl, pH 6.8, 0.1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride), sonicated at the maximum intensity, and pulsed at 70% intervals (Sonicator model W-375, Ultrasonics Inc.) for 5 min in an ice bath. The supernatant from centrifugation at 15,000 ϫ g for 1 h was loaded on a MonoQ column (HR 10/10). After washing the column with buffer A, the proteins were eluted with a KCl gradient (0 -0.6 M KCl in buffer A, over 60 min). Fractions of 2 ml were collected, and the proteins were analyzed by SDS-PAGE and assayed for activity. For the mutant enzymes with little or no deoxyhypusine synthetic activity, K329R, K329A, K287R, and K287A, extreme precautions were taken to prevent any cross-contamination with the wild type or any of the other mutant proteins. These precautions included thorough washing of the sonication probe, strict separation of samples, and purification of each mutant protein on an individually prepared column of new Q-Sepharose (Fastflow) resin in place of the MonoQ column. The clarified cell lysate (5 ml) prepared as described above was applied to a column (ϳ10 ml) of Q-Sepharose that had been equilibrated in buffer A. After washing the column with buffer A, the mutant protein was eluted by the stepwise addition of buffer A containing 0.2, 0.3, 0.4, or 0.5 M NaCl, and 1-ml fractions were collected and a small portion analyzed by SDS-PAGE. The 41-kDa enzyme or mutant protein eluted in 0.3-0.4 M NaCl. The appropriate fractions were pooled, concentrated, and desalted by centrifugation in Centricon 30 filter devices (Amicon). A culture of a transformant with the pET-11a vector without any insert DNA was processed in parallel as a negative control, and bacterial proteins eluting at the same ionic strength (0.3-0.4 M NaCl) as the enzyme were pooled and concentrated. Total protein concentration was determined by the bicinchoninic acid method using reagents and protocol from Pierce, with bovine serum albumin (Pierce) as a standard, and by measuring the SCHEME 1. Deoxyhypusine synthase reaction. Bold arrows indicate the normal pathway of deoxyhypusine synthesis; thin arrows indicate spermidine cleavage in the absence of the protein substrate (eIF-5A precursor); dashed arrows indicate hypothetical spermidine cleavage that bypasses the enzyme intermediate.

absorption at 280 and 260 nm.
In Vitro Assay of Deoxyhypusine Synthesis Activity-The enzyme activity was measured as described previously (11,19,20). A typical reaction mixture contained, in a total volume of 20 l, 0.2 M glycine-NaOH buffer, pH 9.5, 1 mM dithiothreitol, 5 g of bovine serum albumin, 1 mM NAD, 7-9 M (2-5 Ci) [1, H]spermidine, 10 M ec-eIF-5A, and wild type or mutant enzyme protein (0.01-20 g, 0.01-30 units). Incubations were at 37°C for 60 min unless noted otherwise. After trichloroacetic acid precipitation, the precipitate was washed three times with 10% trichloroacetic acid containing putrescine, spermidine, and spermine (1 mM each) dissolved in 0.4 ml of 6 N HCl and hydrolyzed at 105°C for 16 h. The [ 3 H]deoxyhypusine formed was measured after its separation from other components of the acid hydrolyzed protein fraction by ion exchange chromatography, as described previously (19). One unit of activity is defined as the amount of enzyme catalyzing the formation of 1 pmol h Ϫ1 of deoxyhypusine. The calculation of the specific activity was based on the enzyme or mutant protein concentration estimated by the densitometric analysis of Coomassie Blue-stained bands on SDS-PAGE and by the bicinchoninic acid method noted above.
Spermidine Cleavage-Spermidine cleavage activity of the enzyme was measured in the absence of eIF-5A precursor as described previously (19) and as noted in the legend to Table II. After incubation of the reaction mixture at 37°C for the specified time, protein was removed by precipitation with 10% trichloroacetic acid after addition of carrier bovine serum albumin, and the radioactivity in the cleavage products from [1, (19,20). Because the amount of ⌬ 1 -pyrroline was usually less than that of 1,3-diaminopropane (19,20), the quantity of diaminopropane produced was used as a measure of spermidine cleavage.
Enzyme Intermediate Formation-Labeling of the enzyme by [ 3 H]spermidine was carried after NaBH 3 CN reduction as described previously (20) and in the legend to Fig. 2. The degree of labeling in [ 3 H]deoxyhypusine was measured after trichloroacetic acid precipitation of the reduced enzyme intermediate, acid hydrolysis, and ion exchange chromatography of the acid hydrolysate as in the deoxyhypusine synthesis assay (above).

RESULTS
The mutant proteins were expressed in E. coli at approximately the same level as the wild type enzyme. Partially purified preparations (10 -40% purity, Fig. 1A) of the mutant proteins were obtained by one-step ion exchange chromatography.
In each case, the elution pattern from the ion exchange column of the mutant protein (not shown) was quite similar to that of the wild type enzyme. Thus it appears that the mutations did not cause gross changes in conformation or stability of the proteins.
Tetramer Formation by Mutant Enzymes-Deoxyhypusine synthase exists as a homotetramer of 41-kDa subunits in the native state, as evidenced by sedimentation rates upon ultracentrifugation, mobility in gel filtration, and migration upon electrophoresis in nondenaturing gels (11, 12, 16 -18). This structure appears to be important for stability, for binding of substrates, and for catalysis. In fact, a recent x-ray crystallographic study reveals a tetrameric organization of enzyme subunits. 3 To determine if any of the mutations altered the ability of the proteins to form the tetramer, we compared the mobilities of the mutant enzymes with wild type enzyme in nondenaturing as well as SDS gels (Fig. 1). The 41-kDa subunit of all the mutant enzymes migrated the same distance as the wild type enzyme upon SDS-PAGE (Fig. 1A). Furthermore, the pattern of migration on nondenaturing gels (Fig. 1B) suggests that each of the mutant enzymes exists in tetrameric form, similar to that of the wild type enzyme. A very slight difference in mobility on a nondenaturing gel was observed with K156R and K287R, compared with wild type or other mutant proteins (Fig. 1B).
Enzyme-Imine Intermediate Formation-Our initial interest was to identify the lysine residue uniquely involved in the enzyme intermediate formation.   (11). b The mutation at a specific residue number is indicated by one-letter amino acid abbreviations. The first letter is the wild type residue, and the last letter is the amino acid to which it is changed by the mutated codon. mutagenesis ( Table I). The resulting mutant proteins were examined for their ability to form the enzyme intermediate. Fig. 2 shows the pattern of radiolabeling of wild type and mutant enzyme proteins after their incubation with [1, H]spermidine and NAD in the absence of the protein substrate, eIF-5A precursor, followed by NaBH 3 CN reduction. Using approximately the same amount of each enzyme protein (ϳ2 g), intense labeling was observed with the mutant proteins K141R, K156R, K205R, K212R, K226R, and K251R as well as with the wild type. Radiolabeling of K338R was reduced to ϳ38% of the wild type. The labeling of the mutant K287R was too low to be clearly visible at the exposure time of the fluorogram of Fig. 2B. However, when a large amount (50 g) was used, the incorporation of radioactivity into the mutant protein was clearly detectable and was estimated to be ϳ0.7% of wild type enzyme (Table II). A similarly low but definite labeling (0.6% of wild type enzyme) was also observed with the K287A mutant protein (Table II). In contrast, no radiolabeling was detectable (Table II) with K329R or with K329A even when the incubation was scaled up with 130 g of mutant protein.
(With 130 g of wild type enzyme the level of labeling would be ϳ10 ϫ 10 6 dpm.) Radiolabeled deoxyhypusine was released upon acid hydrolysis from all the proteins that were labeled. Upon digestion of the labeled mutant enzymes with trypsin and chymotrypsin only a single labeled peptide was detected in each case, as shown for K287R (Fig. 3). The labeled peptide from each of the mutant proteins showed the same mobility on a two-dimensional peptide map (Fig. 3) as the peptide (identified previously as Gly-deoxyhypusine-Ile-Arg (20)) corresponding to Gly 328 -Arg 331 of the labeled wild type enzyme. These findings indicate that in the mutant proteins, as well as in the wild type enzyme, the enzyme intermediate is formed exclusively with the lysine at residue 329. Human deoxyhypusine synthase wild type enzyme was purified as described previously (11,20). Partially purified mutant enzymes after MonoQ or Q-Sepharose ion exchange chromatography (as described under "Experimental Procedures") (ϳ5 g) was used for SDS-PAGE on a 10% Tricine gel (Novex) under denaturing conditions (A) or (ϳ2 g) for native gel electrophoresis on a 10% Tris-glycine gel under nondenaturing conditions (B), followed by Coomassie Blue staining. The positions of the wild type and mutant enzymes are indicated by arrows. Upon SDS-PAGE on 10% Tricine gels both the wild type and mutant enzymes migrate as a fine doublet band of ϳ41 kDa. The basis for resolution as a doublet is unknown. WT, wild type.

FIG. 2. Comparison of the abilities of mutant enzymes to form an enzyme intermediate in the absence of eIF-5A precursor. A,
Coomassie Blue-stained gel. B, fluorogram. Purified wild type deoxyhypusine synthase or partially purified mutant enzyme (ϳ2 g/ml reaction mixture) was incubated in the absence of eIF-5A precursor as described under "Experimental Procedures" with 7.5 M [1,8-3 H] spermidine (4 Ci/20 l), 500 M NAD, 5% glycerol. After a 2-min incubation at 37°C, the tubes were placed on ice and 0.1 M NaBH 3 CN was added in four portions over 10-min intervals to a final concentration of 20 mM. A portion equal to 25% of the total was removed for trichloroacetic acid precipitation and analysis of the labeled deoxyhypusine in the acid hydrolysate. The proteins in the remaining 75% of the reaction mixture were precipitated with 10% trichloroacetic acid after addition of 25 g of bovine serum albumin; the precipitates were redissolved in 16 l of Tricine sample buffer containing mercaptoethanol, boiled for 3-4 min, and subjected to SDS-PAGE on a 10% Tricine gel (Novex). The gel was stained in Coomassie Blue, destained, soaked in 1 M sodium salicylate for 60 min, and exposed to x-ray film (Kodak, X-OMat AR) for 5 h at Ϫ70°C. WT, wild type. simination to form the enzyme-imine intermediate, (III) transimination to form the eIF-5A-imine intermediate, and (IV) enzyme-coupled reduction of the eIF-5A imine intermediate. Thus, substitution of any one lysine residue of the enzyme could potentially influence one or more aspects of substrate binding, and/or sequential reaction steps. It is not possible to measure each of the four steps above separately. However, one can independently measure the enzyme intermediate formation (step I ϩ II), spermidine cleavage (which requires step I and normally involves step II in the absence of eIF-5A precursor), or deoxyhypusine synthesis (which covers all four steps in the presence of the protein substrate). All three of these reactions start with step I and thereby are dependent on NAD. Certain mutations may exert differential effects on the three aspects of enzymatic activity. We initially introduced Arg in place of Lys to maintain a positive charge at the same site. However, the bulky side chain of arginine in place of the lysine residue may impose local steric hindrance or strain. For the two sites, Lys 287 and Lys 329 , where the Lys 3 Arg substitution caused a marked reduction in activity, Lys to Ala mutations were also introduced to assess whether these lysine residues play an intrinsic role in catalysis.
Lys to Arg substitutions at five residues, 156, 205, 212, 226, and 251, had relatively insignificant effects on the overall activities (Fig. 4). These mutants exhibited enzyme intermediate formation at a rate similar to that of the wild type. Moderate reductions in deoxyhypusine synthesis and spermidine cleavage were observed with some of these proteins.
The mutant protein K141R showed a small reduction in enzyme intermediate formation (Fig. 2) and spermidine cleavage (Fig. 4); its activity in deoxyhypusine synthesis was ϳ20% of the wild type enzyme. Thus it appears that K141R is less efficient than the wild type enzyme in its catalysis of the later steps (step III or IV; Scheme 1). For the mutant protein K338R, spermidine cleavage activity and enzyme intermediate formation were 35 and 38% of the wild type enzyme, respectively, but its deoxyhypusine synthesis activity was reduced further, to ϳ10% of the wild type, suggesting that the K338R mutation affects both the early as well as the late steps of deoxyhypusine synthesis.
A remarkable loss of activity was observed with mutant proteins that have an Arg or Ala substitution at Lys 287 . Activities of these mutant proteins had values too low to distinguish on the bar graph ( Fig. 4) but are listed in Table II. The K287R mutant almost totally lost the ability to form the enzyme intermediate and to catalyze deoxyhypusine synthesis (ϳ0.7 and 1.1% of the wild type level, respectively). K287A also showed very low enzyme labeling (ϳ0.6% of the wild type), but its activity in deoxyhypusine synthesis was even lower (0.03% of the wild type). For both mutant proteins K287R and K287A, spermidine cleavage activities (5.3 and 2.4% of wild type) were higher than their respective deoxyhypusine synthesis activities.
The most striking effects were seen with mutations at Lys 329 . With K329A and K329R, which cannot form the enzyme intermediate ( Fig. 2 and Table II), absolutely no deoxyhypusine synthesis activity could be detected (Table II)   Comparison of specific activities of deoxyhypusine synthase mutants for deoxyhypusine synthesis in eIF-5A, for spermidine cleavage, and for enzyme labeling Specific conditions for the assays are given under "Experimental Procedures." The incubation times for deoxyhypusine synthesis in eIF-5A, spermidine cleavage, and enzyme labeling were 1 h, 30 min, and 2 min, respectively. 100% equals the specific activities of the wild type enzyme under the conditions of these experiments, i.e. formation of deoxyhypusine in eIF-5A, 1100 pmol (ϳ1.8 ϫ 10 7 dpm)/h/g; spermidine cleavage, 27 pmol (ϳ4.6 ϫ 10 5 dpm)/30 min/g; enzyme labeling, 4.5 pmol (ϳ7.6 ϫ 10 4 dpm)/2 min/g. The amount of wild type enzyme used in a typical assay was 0.01, 0.4, and 2 g, respectively, for deoxyhypusine synthesis, spermidine cleavage, and enzyme labeling; the amount of the mutant enzymes was increased up to 20, 43, and 130 g for the three assays in order to obtain a measurable amount of labeled product or to ensure there was no activity. Quantitation was based on the measurement of radioactivity in the product after ion exchange chromatographic separation of the trichloroacetic acid supernatant (for spermidine cleavage) or acid hydrolysates of the labeled proteins (for deoxyhypusine synthesis in eIF-5A and enzyme labeling). The minimum reliably detectable radioactivity in each peak area was 100 dpm. a These values are below the detection limit (Ͻ0.001%).
FIG. 3. Two-dimensional peptide maps prepared from a trypsin ؉ chymotrypsin digest of wild type (A) or mutant K287R (B) enzyme after radiolabeling. Radiolabeling of the enzymes was carried out as described previously (20), using 2 g of the wild type enzyme or 31 g of K287R mutant enzyme. The radiolabeled enzyme band was excised from the SDS gel and thoroughly washed to remove SDS. The labeled protein in the gel slice was digested with trypsin (Worthington, tosylphenylalanyl chloromethyl ketone-treated, 20 g/ml) and chymotrypsin (Worthington, N ␣ -p-tosyl-L-lysine chloromethyl ketone-treated, 120 g/ml) in 0.15 ml of 50 mM ammonium bicarbonate buffer, pH 8.0, for 22 h. The lyophilized digest was separated on silica gel-coated plastic sheets (EM Science) as described (20). DNP, dinitrophenyl.  Table II. WT, wild type. mined by amino acid sequencing (20). Together these studies provide definitive evidence that Lys 329 is the sole site of the covalent enzyme-substrate intermediate formation in human deoxyhypusine synthase and that the enzyme-substrate complex at Lys 329 is an obligatory catalytic intermediate in the course of deoxyhypusine synthesis. Furthermore, the catalytic properties of some mutant proteins with Lys 3 Ala or Lys 3 Arg substitution offer new insights into structure-function relationships and the complex reaction mechanism of deoxyhypusine synthase (Scheme 1).
Even though the mutant enzymes used in this study were partially pure, the observed activities of deoxyhypusine synthesis and spermidine cleavage must be intrinsic to the recombinant proteins and are not due to contaminating E. coli proteins. E. coli does not contain genes for either deoxyhypusine synthase or eIF-5A and cannot contribute any deoxyhypusine synthetic activity. However, contamination with spermidine cleavage activity from a bacterial source was a concern because certain bacteria, e.g. Serratia marcescens and Micrococcus rubens, are known to contain spermidine dehydrogenase/oxidases that cleave spermidine to ⌬ 1 -pyrroline and diaminopropane in a reaction that requires an electron acceptor other than NAD (23,24). The spermidine cleavage, as well as enzyme intermediate formation and deoxyhypusine synthesis activities, observed in the current experiments was strictly dependent on NAD, supporting the assumption that this cleavage results from catalysis by deoxyhypusine synthase. Furthermore, no spermidine cleavage or deoxyhypusine synthesis activity was detectable from a parallel protein preparation from E. coli transformed with the pET11a vector alone without any insert DNA under the same assay conditions.
Comparison of the activities of all the mutant proteins in enzyme intermediate formation, spermidine cleavage, and deoxyhypusine synthesis (Fig. 4) shows a pattern of general correlation between the three aspects of enzyme activity. This would be expected if spermidine cleavage and deoxyhypusine synthesis occur by way of the enzyme intermediate, as proposed in Scheme 1. Defects in enzyme intermediate formation, observed in several of these mutants, caused an almost equal or greater reduction in deoxyhypusine synthesis. Spermidine cleavage activity seemed to correlate generally with enzyme intermediate formation. In the case of the K287R, K287A, and K329A mutants, the relative activity of spermidine cleavage is greater than that of enzyme intermediate formation, suggesting that in special circumstances, spermidine cleavage can occur independently of the formation of the enzyme intermediate, as will be discussed below.
The mutant proteins K329A and K329R offered a key opportunity to assess the role of Lys 329 in enzyme intermediate formation and in the three aspects of catalysis. Even with the biochemical evidence for the enzyme intermediate at Lys 329 in vitro presented in the preceding paper (20), in the case of the mutant enzymes lacking this residue, K329A and/or K329R, it was conceivable that an alternate lysine residue near the active site could be recruited for enzyme intermediate formation.
However, no labeling of the enzyme was observed with these two mutant enzymes, indicating that there is a stringent requirement for the orientation of Lys 329 for it to carry out the transimination involving dehydrospermidine and to function as an acceptor of the butylamine moiety (Scheme 1) (20). The total lack of deoxyhypusine synthesis activity of the K329R and K329A mutant enzymes leads us to conclude that the enzyme intermediate at Lys 329 is critical for the catalysis as the mediator of butylamine transfer to the eIF-5A precursor.
Based on previous studies it was proposed that spermidine cleavage by deoxyhypusine synthase in the absence of the eIF-5A precursor involves the nucleophilic attack by the ⑀-amino group of Lys 329 of the enzyme on the carbon of the NϭC bond of dehydrospermidine, resulting in the release of diaminopropane with concomitant formation of an enzyme-imine intermediate (20). The butylamine side chain of this intermediate, in turn, can undergo cyclization to release ⌬ 1 -pyrroline (Scheme 1, plain arrow) in the absence of eIF-5A precursor, the ultimate acceptor of the four carbon moiety. With K329A, where Lys 329 is missing, NAD-dependent cleavage of spermidine was still detectable (5.8% of wild type enzyme). In this case, spermidine cleavage and cyclization of the butylamine moiety to ⌬ 1 -pyrroline could occur by the nucleophilic attack by the terminal amino group of dehydrospermidine on the C of its own NϭC bond (19,25,26) (see Scheme 1, dashed arrow). In contrast to K329A, no spermidine cleavage activity was detectable with K329R. The difference in spermidine cleavage activities between the two mutant proteins may reflect an interesting feature of the spermidine-binding pocket of the enzyme. Previous studies, in which the spermidine-binding site of the enzyme was probed by the use of various spermidine analogues as inhibitors (27), suggested that the spermidine-binding site is a narrow groove, especially around the secondary nitrogen. It is tempting to speculate that replacement of Lys 329 with a bulky Arg residue introduces steric hindrance that prevents spermidine binding or the cyclization of the butylamine moiety of dehydrospermidine, whereas substitution with Ala does not.
The catalytic properties of the mutant proteins permit us to make certain predictions on the role of these residues in substrate binding, catalytic reactions, and the conformational integrity of the enzyme. The interpretations drawn from these findings are consistent with x-ray crystallographic data obtained for human deoxyhypusine synthase in a complex with NAD. 3 Amino acid residues in contact with NAD were identified and residues involved in spermidine binding can be predicted from the crystallographic data and molecular modeling. 5 Because the Lys to Arg substitution at several sites, i.e. 156, 205, 212, 226, and 251, did not cause a significant change in enzyme intermediate formation, spermidine cleavage, or deoxyhypusine synthesis in the eIF-5A precursor, these residues probably are not directly involved in tetramer assembly, substrate binding, or catalysis. Indeed, residues 141, 156, 205, 212, 226, 251, and 338 are not located near the NAD-binding site or the predicted spermidine-binding site at the active center of the enzyme, and the mutations at some of these sites may influence the activity indirectly by causing a slight alteration in the binding of substrate or in enzyme conformation. Replacement of Lys 287 with either Arg or Ala caused a drastic reduction in the enzyme activities as measured by spermidine cleavage, enzyme intermediate formation, and deoxyhypusine synthesis. Unlike Lys 329 , this residue (Lys 287 ) does not participate directly in the catalytic reaction, nor does it appear to serve in the binding of NAD or spermidine. Lys 287 is next to His 288 , which is predicted to play a key role in the NAD-dependent dehydrogenation of spermidine. 3 It is possible that substitution of Lys 287 with either Ala or Arg could cause a distortion in the orientation of His 288 or in the conformation of the active site such that the ability of the enzyme to interact with spermidine and NAD and/or to catalyze the NAD-dependent dehydrogenation would be seriously compromised.
Most importantly, Lys 329 is located at the active center of the enzyme near the NAD-binding site and in the middle of the predicted spermidine-binding pocket in an orientation that would enable its role as the acceptor of the butylamine moiety from the dehydrospermidine intermediate. It is not one of the anchoring sites for NAD or spermidine. The K329R mutant enzyme forms a tetramer and also can associate with the eIF-5A precursor protein. 6 Judging from the spermidine cleavage activity of K329A, it is clear that the K329A mutant enzyme can carry out NAD-dependent dehydrogenation of spermidine. The lack of deoxyhypusine synthesis activity by both the K329A and K329R mutant enzymes is most likely attributable to their inability to form the enzyme-imine intermediate, which is essential for the butylamine transfer from dehydrospermidine to the eIF-5A precursor. X-ray crystallographic studies of the mutant enzymes should provide further insights into the mechanism and the structure-function relationships of this unusual enzyme.