Enzyme-Substrate Intermediate Formation at Lysine 329 of Human Deoxyhypusine Synthase*

Deoxyhypusine (N ε-(4-aminobutyl)lysine) is the key intermediate in the posttranslational synthesis of the unique amino acid, hypusine (N ε-(4-amino-2-hydroxybutyl)lysine). Deoxyhypusine synthase catalyzes the formation of deoxyhypusine by conjugation of the butylamine moiety of spermidine to the ε-amino group of one specific lysine residue of the eukaryotic translation initiation factor 5A (eIF-5A) precursor protein. However, in the absence of the eIF-5A precursor, catalysis involves only the NAD-dependent cleavage of spermidine to generate 1,3-diaminopropane and a putative 4-carbon amine intermediate that gives rise to Δ1-pyrroline. We have obtained evidence for a covalent enzyme-substrate intermediate that accumulates in the absence of the eIF-5A precursor. Incubation of human recombinant enzyme with [1,8-3H]spermidine and NAD, followed by reduction with NaBH3CN, resulted in specific radiolabeling of the enzyme. The radioactive component in the reduced enzyme intermediate was identified as deoxyhypusine and was shown to occur at a single locus. The fact that labeled deoxyhypusine was found after treatment with a reducing agent suggests an intermediate with the butylamine moiety derived from spermidine attached through an imine linkage to the ε-amino group of a specific lysine residue of the enzyme. This residue has been identified as lysine 329. Separate experiments showing efficient transfer of labeled butylamine moiety from enzyme intermediate to eIF-5A precursor strongly support a reaction mechanism involving an imine intermediate.

The posttranslational formation of deoxyhypusine (N ⑀ -(4aminobutyl)lysine) is catalyzed by deoxyhypusine synthase and involves modification of a single lysine residue (Lys 50 in the human protein) of the eIF-5A 1 precursor (for reviews, see Refs. [1][2][3]. It is the first step in the biosynthesis of N ⑀ -(4-amino-2-hydroxybutyl)lysine (hypusine), the unusual amino acid that characterizes mature eIF-5A (4,5). This 17-kDa protein and the enzymes that catalyze its modification are present in all eukaryotic cells. There is mounting evidence that hypusine synthesis is vital for proliferation in mammalian cells (2, 6 -8) and those of other eukaryotic species. Yeast cells are not viable if the two genes for eIF-5A have been inactivated, whereas expression of either one permits growth (9 -12). Furthermore, it has been shown very recently that the presence of an intact gene for deoxyhypusine synthase (13) 2 and expression of deoxyhypusine synthase activity 2 are essential for yeast viability. Effective inhibitors of deoxyhypusine synthase have been developed; these inhibitors cause inhibition of hypusine synthesis in eIF-5A in cells and also have been shown to inhibit proliferation of various mammalian cells (6,14,15). In view of the position of the enzyme as the first step in the biosynthesis of hypusine, such inhibitors could have an important place in therapy of hyperproliferative conditions. A thorough understanding of the mechanism and regulation of this enzyme is therefore of great interest.
Deoxyhypusine synthase has been purified from rat testis (16), Neurospora crassa (17), and HeLa cells (18). The gene was identified in yeast (13, 18 -20), and human and Neurospora cDNAs have been cloned (20 -22). The recombinant enzymes, purified after expression of the human or yeast cDNA in Escherichia coli, are now available in essentially homogeneous form. The amino acid sequence of the enzyme is highly conserved; it appears that the enzyme from all species studied exists as a tetramer of identical subunits of 40 -43 kDa (depending on the species) and that the cofactor requirements and catalytic properties of the enzyme from different species are similar (16,19,21).
Earlier studies with the purified enzyme from rat testis (16,23), and with the human and yeast recombinant enzymes (19,21), have shown that the complex reaction leading to the modification of the eIF-5A precursor protein through deoxyhypusine synthesis is carried out by a single enzyme. In a complete reaction mixture containing spermidine, NAD, and eIF-5A precursor, deoxyhypusine synthase catalyzes an NADdependent cleavage of spermidine to generate free 1,3-diaminopropane and an enzyme-bound 4-aminobutyl moiety, which is subsequently transferred to a lysine of the eIF-5A precursor to form the deoxyhypusine residue. However, if the eIF-5A precursor is absent from the reaction mixture, deoxyhypusine synthase catalyzes only part of this reaction, namely the cleavage of spermidine; in this case 1,3-diaminopropane and ⌬ 1pyrroline are observed as products (16,19,21,23). The role of dehydrogenation of spermidine was established in experiments that showed the transfer of 3 H from [5-3 H]spermidine to NAD as a critical initial step of the reaction (23). Based on these and other early observations, which established the origin of the atoms of deoxyhypusine (24 -26), we proposed an overall scheme, similar to that shown in Scheme 1. The proposed mechanism involved a dehydrospermidine intermediate, probably tightly bound to the enzyme (23), and possibly an enzyme-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Methods
Peptide Synthesis-The synthetic peptides, Gly-deoxyhypusine (Dhp)-Ile-Arg and Gly-Dhp-Glu-Leu-Arg, was prepared by a solid phase procedure (36) with the use of t-Boc-Arg(Tos)OCH 2 PAM resin (Applied Biosystems) and tert-butyloxycarbonyl ␣-amino group protection. The amino-protected deoxyhypusine derivative, tert-butyloxycarbonyl-N 6 -pnitrobenzyloxycarbonyl-N 6 -(4-p-nitrobenzyloxycarbonylaminobutyl)-(2S)-2,6-diaminohexanoic acid, was prepared as outlined (37). t-Boc-Glu-␥-benzyl ester was used in the synthesis of the latter peptide. The peptides were partially deprotected and removed from the resin with HF containing 10% m-cresol. Removal of the p-nitrobenzyloxycarbonyl groups from the side chain of the deoxyhypusine residue was then carried out by hydrogenation in 50% aqueous methanol in the presence of palladium black catalyst. The peptides were purified by chromatography on CM-Sephadex C-25 (NH 4 ϩ form, Pharmacia) using a linear gradient of increasing strength of NH 4 OH. Examination of acid hydrolysates of the peptides by thin layer chromatography showed, in each case, approximately equal amounts of each of the composite amino acids.
Purification of Human Deoxyhypusine Synthase-Purification of recombinant enzyme from E. coli B strain BL21 (DE23) expressing human deoxyhypusine synthase cDNA followed the published procedure (21) with modifications. Because the enzyme prepared as described earlier (21) that appeared homogeneous on SDS-PAGE in a Tris-glycine buffer system in fact contained a significant amount of a bacterial protein of nearly the same molecular mass, as became apparent upon SDS-PAGE in a Tricine buffer system, an additional step was taken. To separate the enzyme from the contaminating bacterial protein, chromatography was conducted on a butyl-Sepharose column. The sample after Mono-Q chromatography (16) was applied to this column equilibrated with 0.05 M Tris acetate, pH 7.5, containing 0.1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 1 M (NH 4 ) 2 SO 4 . Elution was made with an 80-ml linear gradient from equilibration buffer to the same buffer without (NH 4 ) 2 SO 4 . The enzyme activity eluted just after the conductivity dropped to ϳ40% of the initial value. This preparation, estimated to be Ͼ95% pure (specific activity 900 units/g, where 1 unit ϭ formation of 1 pmol of deoxyhypusine/h) was used for the experiments reported in this study. The enzyme activity was assayed by the published procedure (16,23).  (23). Abbreviations are as follows: Pre, eIF-5A precursor protein; Enz, deoxyhypusine synthase; Lys-NH 2 , schematic representation of the lysine that is modified in the eIF-5A precursor or in the enzyme, to show only the ⑀-amino group; Enz(Dhp), modified enzyme containing deoxyhypusine.
Generation of an Enzyme-Substrate Intermediate and Its Reduction by NaBH 3 CN-Highly purified human deoxyhypusine synthase (1-3 l, 2-6 g) was added to mixtures of reactants preheated at 37°C in siliconized Eppendorf tubes to give, typically, 20 l containing 1 mM dithiothreitol, 5% glycerol, 500 M NAD, and 7.5 M [1,8-3 H]spermidine (4 Ci) in 0.2 M glycine-NaOH buffer, pH 9.5. After 2 min, or the specified time, at 37°C, the tubes were transferred to an ice bath and NaBH 3 CN was added (1 l of 0.1 M in H 2 O, freshly prepared from the recrystallized compound, for a total of 3-4 portions at 10-min intervals) to a final concentration of 15-20 mM. Specific details are given in the figure legends. For SDS-PAGE, 25-100 g of bovine serum albumin was added to a portion of each reaction mixture and the proteins were precipitated with 10% trichloroacetic acid. The precipitates were dissolved in Tricine sample buffer containing mercaptoethanol, adjusted to neutral pH with Tris base, placed in boiling water for 3-4 min, and subjected to SDS-PAGE on a precast 10% (Novex) Tricine gel. The gel was stained with Coomassie Blue, destained, soaked in water for 30 min and in 1 M sodium salicylate for 60 min, and exposed to Kodak X-Omat AR x-ray film at Ϫ70°C. For ion-exchange chromatographic analysis of deoxyhypusine and polyamines, another portion of the enzyme-labeling reaction mixture was precipitated with 10% trichloroacetic acid in the presence of 500 g of carrier BSA, the precipitates washed three or four times with 10% trichloroacetic acid containing 2 mM spermidine, and subjected to acid hydrolysis in 6 N HCl at 105°C for 16 h. The hydrolysate and the trichloroacetic acid-soluble reaction products were analyzed by an ion-exchange chromatography system described previously (38) employing sodium citrate/NaCl buffers: Buffer A (0.6 N Na ϩ , pH 5.8), Buffer B (1.5 N Na ϩ , pH 5.55), Buffer C (3 N Na ϩ , pH 5.55) for 5, 10, and 30 min, respectively. The identification of reaction products was as reported previously (23).
Isolation of Peptides and Sequence Determination-For sequence determination, labeled human deoxyhypusine synthase was prepared as outlined above on a large scale, using 400 g of enzyme in a final volume of 2 ml; [1, H]spermidine was diluted with non-radioactive spermidine to a final concentration of 108 M (400 Ci/2 ml). After reduction of the enzyme intermediate, protein was precipitated with trichloroacetic acid without addition of bovine serum albumin, and the precipitates were washed thoroughly with 10% trichloroacetic acid containing 2 mM spermidine, followed by ice-cold acetone (two washes). The protein was digested for 10 h at 37°C in 100 mM ammonium bicarbonate, pH 8.1, containing 0.15 units of trypsin (300 units/mg) and 0.1% reduced Triton (Sigma). The trypsin level was brought to 0.3 units, and the digestion was continued for an additional 14 h. Separation of peptides was performed on a 2 mm ϫ 250 mm C-18 column (Vydac) in 0.1% trifluoroacetic acid in HPLC water; a gradient was then applied over 90 min bringing the elution buffer (90% acetonitrile, 10% HPLC water, 0.085% trifluoroacetic acid) to 65% acetonitrile. A hydrophobic radiolabeled peptide eluting as a single peak later in the gradient was identified as peptide A by electrophoresis and TLC as described in the legend to Fig. 5. Another radiolabeled peptide eluted from the C-18 early in the gradient was identified as peptide B; after further purification on an ion-exchange column, the radioactive peak fraction was desalted using a Perkin-Elmer Applied Biosystems C-18 column, 1.0 mm ϫ 50 mm, in 0.1% trifluoroacetic acid, brought to 50% acetonitrile:50% water over 30 min. This peptide was subjected to 7 cycles of Edman automated degradation to yield GXIR as a major sequence; no equivalent amount of amino acid derivative was seen after the fourth cycle. The tryptic digestion, separation of peptides, and sequence determination by automated Edman degradation were carried out at the Michigan State Macromolecular Structure Facility under the direction of Joseph F. Leykam, Department of Biochemistry, Michigan State University, East Lansing, MI.

RESULTS
Trapping of an Unsaturated Enzyme-Substrate Intermediate-It was observed previously that deoxyhypusine synthase acts to cleave spermidine to 1,3-diaminopropane and a 4-carbon amine moiety both in the presence and absence of the eIF-5A precursor, although the nature of the 4-carbon product is different in the two cases. The butylamine moiety becomes the extended side chain of deoxyhypusine in the eIF-5A precursor protein in the former case and free ⌬ 1 -pyrroline in the latter (23). We reasoned that if a common, transient intermediate is involved in formation of the two 4-carbon products, it would be more likely to accumulate in the absence of eIF-5A precursor, since in this case the spermidine cleavage reaction proceeds at a much slower rate. If, after the initial dehydrogenation of spermidine and cleavage of the N4 -C5 bond, the resulting butylamine moiety is covalently attached to enzyme through an unsaturated linkage, it should be possible to produce a stable derivative of the postulated intermediate by reduction of the reaction mixture. We succeeded in demonstrating that, in the absence of the normal acceptor of the butylamine moiety, the eIF-5A precursor, radioactivity from [1, H]spermidine was indeed retained on the enzyme protein after brief incubation of deoxyhypusine synthase, NAD and [1, H]spermidine if the reaction mixture was subsequently treated with either NaBH 4 or NaBH 3 CN. The labeling appeared to be more efficient with NaBH 3 CN than NaBH 4 , as noted by others (34,35). The results of a typical experiment are shown in Fig. 1. In the Coomassie Blue-stained gel (Fig. 1A), the single band of human deoxyhypusine synthase is shown at ϳ41 kDa (indicated by the arrow), along with several bands from bovine serum albumin, added to aid in the trichloroacetic acid precipitation of protein before SDS-PAGE. The fluorogram of the same gel (Fig. 1B) shows one band of radioactivity, which is at the position of the enzyme (arrow), only in the sample in which the enzyme was incubated with NAD and [ 3 H]spermidine, followed by reduction with NaBH 3 CN (Fig. 1B, lane 5). No labeled product was detectable if enzyme or NAD was omitted from the reaction mixture (Fig. 1B, lanes 1-3) or when reducing agent was not added (Fig. 1B, lanes 2 and 4).
Identification of Deoxyhypusine as the Product of Reduction of the Enzyme Intermediate-The radioactive component of the 3 H-labeled enzyme band excised from a gel or of the trichloroacetic acid precipitates of a reaction mixture ( Fig. 2A) was identified as deoxyhypusine by ion exchange chromatography after acid hydrolysis (see "Experimental Procedures"). A single peak of radioactivity, corresponding in position to that of deoxyhypusine, was observed ( Fig. 2A, q). No detectable [ 3 H]deoxyhypusine was found in those samples where NAD  4). A portion (75%) of each reaction mixture was subjected to SDS-PAGE on a 10% Tricine gel, and, after fluorographic treatment, exposed to x-ray film for 6 h at Ϫ70°C. was omitted ( Fig. 2A, E), or where NaBH 3 CN was not added (data not shown). In this experiment, during the 2-min incubation, 21.1 pmol of deoxyhypusine/100 pmol of enzyme monomer was formed. Analysis by ion-exchange chromatography of the trichloroacetic acid-soluble radioactive products (Fig. 2B) formed from [1, H]spermidine revealed pyrrolidine (the reduction product of ⌬ 1 -pyrroline; Ref. 23) and 1,3-diaminopropane. Interestingly, the amount of pyrrolidine is significantly less than the amount of 1,3-diaminopropane generated (7.3 versus 30.3 pmol, respectively); the difference can largely be accounted for by the amount of the 4-aminobutyl moiety incorporated into the enzyme and trapped as deoxyhypusine (21.1 pmol) (Fig. 2A). On the other hand, in a parallel reaction not treated with NaBH 3 CN, the ⌬ 1 -pyrroline and 1,3-diaminopropane were found in nearly equal quantities (ratio of 0.8), as observed previously (23). This can be taken as evidence that the enzyme intermediate (Enz-NϭCH(CH 2 ) 3 NH 2 in Scheme 1), if not trapped by NaBH 3 CN reduction, gives rise to ⌬ 1 -pyrroline.
Time Course of Formation of Deoxyhypusine Synthase-Substrate Intermediate-The level of the enzyme intermediate, measured as the amount of deoxyhypusine after reduction (Fig.  3A), reached a peak at ϳ2 min and appeared to decrease slowly thereafter. A parallel increase in 1,3-diaminopropane occurred in the first 2 min (Fig. 3B). However, in a second phase of slower product formation, the amount of 1,3-diaminopropane and ⌬ 1 -pyrroline (measured as pyrrolidine, see previous section) continued to increase with time (Fig. 3B). Reduction in the level of spermidine corresponded to the increase in 1,3-diaminopropane. The sum of the amounts of ⌬ 1 -pyrroline and inter-mediate correlates generally with the amounts of 1,3-diaminopropane, as discussed in the preceding section. The fact that the enzyme-substrate intermediate did not continue to accumulate, while the other products did, suggests that intermediate formation is balanced by its degradation to ⌬ 1 -pyrroline and free enzyme.
Catalytic Competence of the Deoxyhypusine Synthase-Substrate Intermediate-The postulated enzyme-substrate intermediate should be able to transfer its butylamine moiety to the specific lysine of the eIF-5A precursor if it is a true catalytic intermediate in the reaction. The intermediate, however, is not stable, i.e. it breaks down spontaneously in the absence of eIF-5A precursor, as is evident from the results in Fig. 3. Rather than attempting physical isolation of this transient enzyme-substrate intermediate free from [ 3 H]spermidine, we chose to block its further formation with the use of an inhibitor. 1-Amino,7-guanidinoheptane (GC7) by competition effectively prevents the productive binding of spermidine to the enzyme (11). At 100 M this inhibitor totally stopped intermediate formation, as shown in Fig. 4A (lane 5). That the labeled butylamine moiety from the preformed intermediate can be transferred to eIF-5A precursor, and that the transfer is not prevented by the inhibitor is evident from the data of Fig. 4B. Deoxyhypusine synthase was incubated for a short time with [1, H]spermidine and NAD to accumulate the enzyme intermediate. Reduction followed by PAGE shows evidence for labeled enzyme-intermediate formation (Fig. 4B, lane 1 After trichloroacetic acid precipitation of 25% of the reaction mixture ( Fig. 1) in the presence of 500 g of BSA, the precipitates were washed and subjected to acid hydrolysis. Portions of the hydrolysate (A) and the trichloroacetic acid-soluble reaction products (B) were analyzed by ion exchange chromatography as described under "Experimental Procedures." The solid symbols (A, q; B, ࡗ) depict labeled components from the reaction mixture containing NAD (see Fig. 1, lane 5); the open symbols (A, E; B, Ⅺ) those from the mixture without NAD (see Fig. 1,  lane 3). The values were corrected for a low level of background impurities in the [1, H]spermidine and are expressed as total picomoles/20 l of original reaction mixture, containing ϳ100 pmol of enzyme subunit. DAP, 1,3-diaminopropane; Spd, spermidine.

FIG. 3. Time course of formation of enzyme intermediate (A) and soluble spermidine cleavage products (B).
The reaction mixtures were similar to those described in Fig. 1, except that deoxyhypusine synthase was ϳ2 g (ϳ50 pmol)/20 l, and levels of NAD and [1, H]spermidine were 200 M and 10 M, respectively. Reduction with NaBH 3 CN and ion-exchange chromatographic analysis of labeled deoxyhypusine in modified enzyme after acid hydrolysis (A) and of the trichloroacetic acid-soluble reaction products (B) were carried out as described under "Experimental Procedures" and in the legend to Fig. 2. amine moiety from the enzyme intermediate to the eIF-5A precursor had occurred prior to reduction. In a control reaction, in which inhibitor was included from the onset (Fig. 4B, lane 6), no labeling of either enzyme-intermediate or eIF-5A precursor was observed, confirming that intermediate formation was blocked by the inhibitor. The radioactive component of the modified ec-eIF-5A was identified as deoxyhypusine in both reduced and unreduced samples, showing a normal course of deoxyhypusine production from the enzyme-substrate intermediate, i.e. transfer of the butylamine moiety followed by enzyme-catalyzed reduction. A, Fig. 5A). Chymotrypsin treatment of the tryptic digest containing peptide A provided a distinctly different labeled peptide (Peptide B, Fig. 5B). Prolonged tryptic digestion (40 h) or the use of high levels of trypsin alone also produced peptide B, probably due to a chymotryptic activity recognized to be residual in trypsin even after TPCK treatment, that is enhanced by autodigestion (41). Peptide B is clearly more basic than peptide A, judging by its migration in electrophoresis (Fig.  5), and probably very hydrophilic, as it migrated very little in the chromatography mixture used (Fig. 5). In separate experiments, peptide B was shown to be much smaller than peptide A (Ͻ Ͻ2.5 kDa versus ϳ3.5-6 kDa) by SDS-PAGE on a 16% Tricine gel (data not shown). It seems evident from these results that peptide B is derived from peptide A by a chymotryptic cleavage, with the radiolabel residing in peptide B. Fig. 6A shows the pattern of radioactivity obtained upon Edman degradation of labeled peptide B. The large majority of radioactivity appeared at the second Edman cycle. Examination of the deduced amino acid sequence of human deoxyhypusine synthase (21) revealed two locations where a conserved lysine is the second residue after a tryptic or chymotryptic cleavage site and where the expected peptide would have the characteristics of basicity, hydrophilicity, and small size expected from the results described above. These regions of the sequence, residues 155-159 and 328 -331, along with the flanking 3 residues in each case, are shown in Fig. 6B. The sequence, GXIR, where X is an unidentified residue, was found as a major sequence in a parallel Edman degradation of partially purified peptide B; it is evident that the unidentified residue, X, corresponds to the position of radioactivity shown in Fig. 6A, and to the position of Lys 329 in the unmodified human enzyme (Fig.  6B). Chymotryptic-like cleavage at Trp 327 of the larger peptide FIG. 4. Transfer of the butylamine moiety from enzyme-substrate intermediate to eIF-5A precursor. A, inhibition of intermediate formation by 1-amino-7-guanidinoheptane. Deoxyhypusine synthase samples (ϳ2 g, ϳ50 pmol of monomer/20 l) were incubated as described under "Experimental Procedures" except for 3 min at 37°C and in the presence of the concentrations of inhibitor indicated below, prior to reduction with NaBH 3 CN. To one-half of each reduced sample was added 25 g of bovine serum albumin, and the proteins were precipitated with trichloroacetic acid and subjected to SDS-PAGE on 14% Tris-glycine gels (Novex). Only the fluorogram is shown. Lanes 1-5, 0, 10, 20, 50, and 100 M inhibitor, respectively. B, labeling of eIF-5A precursor. Deoxyhypusine synthase (7 g/88 l) was incubated as in A without inhibitor. After 3 min, inhibitor was added to 100 M final inhibitor concentration. After an additional 3 min, ec-eIF-5A was added to ϳ13.5 M final concentration. At each of the indicated times, a 20-l portion was removed. A 5-l portion of each reaction mixture was prepared for SDS-PAGE with reduction (lanes 1-6) and an equal portion without reduction (lanes 7-10); the remainder was used for deoxyhypusine analysis. Lane 1, without ec-eIF-5A addition; lanes 2-5 and 7-10, 1, 5, 10, and 20 min, respectively, after addition of ec-eIF-5A; lane 6, inhibitor at 100 M added at the onset of incubation, in a separate 20-l reaction. In these electrophoresis conditions, the modified protein (eIF-5A(Dhp) migrates exactly as ec-eIF-5A.  (39). The standards, ⑀-2,4-dinitrophenyl (DNP)-L-lysine and mono-5-dimethylaminonaphthalenesulfonyl (dansyl)-cadaverine, were applied with the samples. Fluorograms were prepared after treatment of the maps with En 3 Hance (DuPont NEN) and exposed to Kodak X-Omat AR x-ray film at Ϫ70°C. No difference was observed in mobility of the labeled peptides when the protein in the gel was reduced and alkylated with 4-vinylpyridine according to the procedure of Moritz et al. (40) prior to digestion.

Identification of Lysine 329 as the Enzyme Residue Involved in Intermediate Formation-Human deoxyhypusine synthase labeled with [ 3 H]spermidine was digested with trypsin. A peptide map of the digest displayed a single radiolabeled peptide (Peptide
A would explain the origin of peptide B. Synthetic peptides corresponding to each of these sequences with deoxyhypusine in place of lysine were prepared (see "Experimental Procedures") and subjected to electrophoresis and chromatography, in two solvent systems, along with radiolabeled Peptide B from the digest (above). The synthetic peptide Gly-Dhp-Ile-Arg exactly co-migrated with the radioactivity in peptide B, whereas the other peptide, Gly-Dhp-Glu-Leu-Arg, did not (Fig. 6C). Thus, the deoxyhypusine found in the modified enzyme after reduction of the enzyme-substrate intermediate must result from modification of Lys 329 of the human enzyme. DISCUSSION The present study was undertaken to extend our knowledge of the deoxyhypusine synthase reaction. The expectation that an enzyme-substrate intermediate might form with the 4-carbon amine moiety from spermidine covalently attached to deoxyhypusine synthase through an unsaturated linkage was based on the earlier observation that the dehydrogenation of spermidine between N4 and C5, as shown by the transfer of 3 H from [5-3 H]spermidine to NAD (23), is a critical obligatory step preceding spermidine cleavage and deoxyhypusine synthesis (see Scheme 1). However, it was not known whether the transfer of the butylamine moiety to eIF-5A precursor occurs directly or through a covalent enzyme intermediate. In one possible scenario, the eIF-5A precursor protein could bind to the enzyme in close proximity to spermidine and transfer of the butylamine moiety could proceed directly to the specific lysine residue of the substrate protein. Thus, one function of the enzyme would be to bring the two substrates into juxtaposition for the transfer. In addition, the enzyme might play a more active role by accepting the butylamine moiety from spermidine to form an enzyme-substrate intermediate and then delivering it to the protein substrate. The current data provide compelling evidence for such a covalent intermediate function for the enzyme.
Identification of labeled deoxyhypusine as a component of modified deoxyhypusine synthase after reduction of enzymesubstrate intermediate defines a lysine residue of the enzyme itself as an acceptor of the 4-aminobutyl moiety from spermidine. That the labeled deoxyhypusine residue on the modified enzyme was generated from an unsaturated intermediate by introduction of hydrogen was confirmed with the use of [ 3 H]NaBH 4 . Radiolabeled deoxyhypusine was produced when an incubation mixture of the enzyme, NAD, and unlabeled spermidine were reduced with [ 3 H]NaBH 4 (data not shown).
Examination of amino acid sequences of the labeled peptides derived from the reduced enzyme-substrate intermediate (Fig.  6) revealed that modification occurred at one specific lysine (Lys 329 ) of human deoxyhypusine synthase. No evidence for involvement of any other residue was obtained. The specificity of the transfer exclusively to this lysine suggests that this is not a random event, but rather is a specific transfer near or at the active site of the enzyme, that takes place immediately following the dehydrogenation of spermidine. However, it was not clear initially whether this reflected an obligatory intermediate in the pathway leading to the production of deoxyhypusine in the eIF-5A intermediate form, eIF-5A(Dhp) (Scheme 1), or an abortive side-product resulting from the absence of the normal acceptor of the butylamine moiety, the eIF-5A precursor. Strong evidence in favor of the former is derived from the observed capability of the enzyme-substrate intermediate to complete a cycle in the deoxyhypusine synthesis when eIF-5A precursor protein is added (Fig. 4).
While these studies were in progress, we conducted independent site-directed mutagenesis studies designed to assess the requisite role of several conserved lysine residues in catalysis. A human mutant enzyme with substitution of Ala for Lys 329 did not form an enzyme intermediate that could be trapped by NaBH 3 CN and had only a trace of deoxyhypusine synthesis activity. 3 These findings are consistent with Lys 329 as the site of enzyme intermediate formation and its involvement in catalysis as an active site residue.
The overall scheme for deoxyhypusine synthesis (Scheme 1, solid arrows) postulates three imine-intermediates, formed in sequence. It is possible to write a plausible reaction scheme A, sequencing of labeled peptide B. Labeled peptide B was prepared from labeled enzyme by digestion with trypsin alone for 40 h at 37°C. Its radiopurity was determined by peptide mapping, as described in the legend to Fig. 5. A portion containing ϳ92,000 dpm was subjected to 13 cycles of automated Edman degradation, and the radioactivity in a portion of the eluant after each cycle was measured. B, partial sequence of human deoxyhypusine synthase in two regions of interest. Portions of the deduced amino acid sequence of human deoxyhypusine synthase (21), showing cleavage sites for trypsin (2), and chymotrysin (Ç). The sequence of peptide B was determined by Edman degradation after partial purification (see "Experimental Procedures"). C, comparison of mobility of radiolabeled peptide B with two synthetic peptides, Gly-Dhp-Glu-Leu-Arg and Gly-Dhp-Ile-Arg. A mixture of radiolabeled peptide B and each synthetic peptide was analyzed on the same thin-layer sheets under two sets of conditions: Condition I, the same as shown in Fig. 5, and Condition II, the same electrophoresis conditions as described for Fig. 5, but chromatography in a different solvent mixture, i.e. methylene chloride:methanol:ammonium hydroxide, 2:2:1.
based on an imine transfer mechanism, as has been extensively described by Jencks (42)(43)(44)(45) and demonstrated for enzymes utilizing pyridoxal phosphate as a cofactor (43,45,46). When protonated, the imine group is highly reactive and can undergo transimination (42)(43)(44)(45). In the case of deoxyhypusine synthase, the initial protonated imine could arise through removal of a hydride ion (H Ϫ ) from spermidine. 4 Ensuing transimination involving the initial imine (dehydrospermidine) and the enzyme results in the formation of the next imine intermediate (Enz-Lys-NϭCH(CH 2 ) 3 NH 2 ) with the release of 1,3-diaminopropane. Subsequent transimination(s) transfers the 4-carbon moiety from the unsaturated enzyme intermediate to the eIF-5A precursor, forming an eIF-5A imine intermediate (Pre-Lys-NϭCH(CH 2 ) 3 NH 2 ). This eIF-5A imine intermediate undergoes enzyme-catalyzed reduction to the deoxyhypusine form (eIF-5A(Dhp)) without exogenous reducing agent. The origin of the hydride ion needed for this final step in the catalytic reaction is not known. NADH generated during the dehydrogenation of spermidine is its most likely source, either direct or indirect. On the other hand, NADH does not reduce the unsaturated enzyme intermediate, presumably due to an unfavorable orientation of the enzyme-imine with respect to NADH. The absence of enzyme imine-intermediate reduction precludes autoinactivation of the enzyme.
In the absence of eIF-5A precursor (Scheme 1, dashed arrows), the unsaturated 4-carbon amine moiety would be retained on the enzyme. This imine could slowly undergo cyclization to ⌬ 1 -pyrroline if its own free amino group acts as the entering nucleophilic group to attack the imine at the secondary nitrogen, as suggested previously (23). The time course of spermidine cleavage in the absence of the protein substrate (Fig. 3) shows indication of a delayed appearance of ⌬ 1 -pyrroline relative to 1,3-diaminopropane. This implies that ⌬ 1 -pyrroline is not derived directly from the butylamine of dehydrospermidine by nucleophilic attack on its own imine, but rather from the enzyme-intermediate. It is possibly due to this abortive cleavage that only a portion of the modified enzyme imine intermediate could be trapped by reduction.
We had previously made a systematic study, using a series of spermidine analogs, to probe the active site of deoxyhypusine synthase (14,15). From the observed efficiency of these inhibitors, we concluded that the orientation of the spermidine molecule, and specifically its secondary amine, N4, must be precisely defined in a cleft on the enzyme. The current studies, which indicate a specific relationship between lysine 329 of the human enzyme and N4 -C5 of spermidine, are consistent with that suggestion. Determination of the x-ray crystal structure, currently under way 5 should shed light on the spatial organization of spermidine and NAD at the active sites of the enzyme and permit tentative identification of amino acid residues involved in spermidine binding and catalysis.