Regulation of Biosynthesis of Hypusine in Chinese Hamster Ovary Cells EVIDENCE FOR eIF-4D PRECURSOR POLYPEPTIDES*

The effects of spermidine depletion and the effects of translation inhibition on hypusine biosynthesis were studied in Chinese hamster ovary cells. Upon depletion of cellular spermidine by treatment with DL-a-difluo-romethylornithine for 42 h or longer, both the rate of deoxyhypusine + hypusine synthesis and the content of protein-bound hypusine were significantly reduced. Cycloheximide caused complete inhibition of deoxyhypusine + hypusine synthesis in untreated cells and in cells in which the spermidine level was reduced to approximately 10% that of the untreated cells by incubation with DL-a-difluoromethylornithine for 24 h. In contrast, the initial synthesis of deoxyhypusine + hypusine was not arrested by cycloheximide in cells depleted of spermidine by treatment with DL-a-difluo-romethylornithine for 42 h. The initial rate of deoxyhypusine + hypusine production in these spermidine-depleted cells increased 5- to 10-fold when the cellular spermidine level was restored through addition of this polyamine to the culture medium. These findings suggest that in control Chinese hamster ovary cells and in cells containing -10% of the control level of spermidine, deoxyhypusine + hypusine synthesis occurs during or immediately after eukaryotic initiation factor 4D precursor translation. However, in cells during depletion of spermidine, there is an accumulation of an eukaryotic

The effects of spermidine depletion and the effects of translation inhibition on hypusine biosynthesis were studied in Chinese hamster ovary cells. Upon depletion of cellular spermidine by treatment with DL-a-difluoromethylornithine for 42 h or longer, both the rate of deoxyhypusine + hypusine synthesis and the content of protein-bound hypusine were significantly reduced. Cycloheximide caused complete inhibition of deoxyhypusine + hypusine synthesis in untreated cells and in cells in which the spermidine level was reduced to approximately 10% that of the untreated cells by incubation with DL-a-difluoromethylornithine for 24 h.
In contrast, the initial synthesis of deoxyhypusine + hypusine was not arrested by cycloheximide in cells depleted of spermidine by treatment with DL-a-difluoromethylornithine for 42 h. The initial rate of deoxyhypusine + hypusine production in these spermidinedepleted cells increased 5-to 10-fold when the cellular spermidine level was restored through addition of this polyamine to the culture medium. These findings suggest that in control Chinese hamster ovary cells and in cells containing -10% of the control level of spermidine, deoxyhypusine + hypusine synthesis occurs during or immediately after eukaryotic initiation factor 4D precursor translation. However, in cells during depletion of spermidine, there is an accumulation of an eukaryotic initiation factor 4D precursor that contains no hypusine or deoxyhypusine, and in these cells deoxyhypusine + hypusine synthesis is mainly regulated by the cellular level of spermidine.
Eukaryotic translation initiation factor 4D (eIF-4D)' is the only cellular protein thus far known to contain the unusual amino acid hypusine (N'-(4-amino-2-hydroxybutyl)lysine) (1). The biosynthesis of hypusine occurs by a novel posttranslational event in which the 4-aminobutyl moiety of the polyamine spermidine is transferred to the eamino group of a specific lysine residue of an eIF-4D precursor to form deoxyhypusine and in which the deoxyhypusine residue of this polypeptide is subsequently hydroxylated (2-4). Thus the biogenesis of eIF-4D may be divided into three stages as shown in Scheme 1: 1) Translation of eIF-4D mRNA; 2 ) post-* 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.
The rate of hypusine formation parallels the increase in protein synthesis in mitogen-treated human peripheral lymphocytes ( 5 ) and correlates in general with the rate of cellular proliferation in various mammalian cells (6)(7)(8). The overall regulation of hypusine biosynthesis, although not clearly understood, may depend upon the rate of synthesis of eIF-4D precursor I (Scheme l), the concentration of this precursor, the level of intracellular spermidine, and/or the activities of the enzymes involved.
The level of spermidine in mammalian cells can be manipulated by the use of DFMO (9), which is an irreversible inhibitor of ornithine decarboxylase, a regulatory enzyme in the biosynthesis of polyamines (10). The dependence of the rate of hypusine synthesis on spermidine concentration has been reported in DFMO-treated rat hepatoma (HTC) cells (8,11). Furthermore, the presence of a pool of unmodified eIF-4D precursor in these cells was postulated on the basis of an immediate increase in the rate of hypusine synthesis following restoration of cellular spermidine (8). A similar eIF-4D precursor was also proposed to be present in resting human peripheral lymphocytes ( 5 ) . However, no definitive evidence in support of these proposals has thus far been provided.
The present study was undertaken to gain insight into the overall regulation of this post-translational modification in DFMO-treated and untreated CHO cells. The experiments described here provide evidence that the initial step of posttranslational modification in untreated cells occurs immediately following the translation of eIF-4D mRNA, whereas in spermidine-deprived cells, the post-translational modification (step 1 in Scheme 1) is repressed and eIF-4D precursor I, which contains no hypusine or deoxyhypusine, accumulates. In these cells, which contain a pool of eIF-4D precursor I, spermidine appears to be the major factor governing the rate of deoxyhypusine + hypusine biosynthesis through control of modification step 1 in Scheme 1. Recently Duncan and Hershey (12) proposed a coupled translation/hypusine modification of eIF-4D in HeLa cells. Our observations in CHO cells provide new information on the relationships between the translation and the two steps of post-translational modification and thus contribute a further understanding of the regulation of hypusine synthesis. Ci/mmol) was purchased from DuPont-New England Nuclear; cycloheximide from Sigma. Other materials, reagents, and the general procedures for cell culture, radiolabeling, and ion exchange chroma-

RESULTS
The data presented in Fig. 1 show that the depletion of cellular spermidine by treatment with DFMO causes a reduction in the content of protein-bound hypusine. When CHO cells were treated with a 4 mM level of DFMO, cellular putrescine dropped to below a detectable level within 24 h. Spermidine was reduced to about 10% of that of untreated cells at 24 h and was not detectable after 48 h of treatment with DFMO (Table I). The content of protein-bound hypusine was lowered by -20% after 24 h but dropped to approximately 30% of the value of untreated cells between 48 and 72 h of treatment ( Fig. 1). This reduction in protein-bound hypusine may result from decreased hypusine synthesis, from increased degradation of eIF-4D, or from changes of both. To date, the effect of polyamine depletion on the turnover of eIF-4D has not been precisely determined. However, the marked repres-  Table I. Trichloroacetic acid precipitates of cells from 10-30 dishes were hydrolyzed in 6 N HCl, and hypusine in the digests was partially purified by ion exchange chromatography on Bio-Rex 70 as previously described (4). Hypusine content in the partially purified fraction was determined by ion exchange chromatography and fluorometric detection as described (4, 13).

TABLE I Polyamine content of untreated and DFMO-treated CHO cells
Cells were plated in 100-mm dishes a t approximately 1 X lo6 cells/ dish in the a-modification of Eagle's medium supplemented as described (3). After l day, DFMO (4 mM) was added. At 24, 48, and 72 h after DFMO addition, cells were washed and harvested as described ( 3 ) . To the pellet of cells from one dish, 0.2 ml of 5% trichloroacetic acid was added. The polyamines in 0.1 ml of the acid supernatant were measured with the use of fluorometric detection after their ion exchange chromatographic separation employing the three-buffer system described earlier ( sion of the post-translational modification upon spermidine depletion as shown in Fig. 2C suggests that decreased synthesis may be mainly responsible for the reduction in proteinbound hypusine. The reduction of hypusine was not as pronounced as that of spermidine in these DFMO-treated cells, presumably because only a minute portion of the total cellular spermidine is used for hypusine production and because the half-life of eIF-4D is relatively long (8).
The relationships between cellular spermidine levels and the initial rates of modification (step 1 in Scheme 1) were assessed in untreated cells and in those treated with DFMO (Fig. 2). In newly modified protein, that which contains radioactivity provided through a structural contribution from [3H]spermidine, the transient intermediate ['Hldeoxyhypusine was seen in addition to ['Hlhypusine. Parenthetically, in the digest of total cellular proteins, only hypusine was detected with the use of the fluorometric method (Fig. l)  The intracellular polyamine content, the polyamine specific radioactivities, and the protein-bound radioactivities in hypusine and deoxyhypusine were measured. The levels of newly formed deoxyhypusine + hypusine were estimated on the assumption that the specific radioactivities of these amino acids are directly related to that of cellular spermidine. The specific radioactivity of intracellular spermidine changes continuously after addition of ['Hlspermidine. In the following calculation the spermidine-specific radioactivities employed were those measured at the middle of the incubation periods. The values for newly formed protein-bound deoxyhypusine + hypusine were estimated with the use of the following expression: Newly formed protein-bound deoxyhypusine + hypusine (pmol/mg protein) = cpm (deoxyhypusine + hypusine)/(% X SRA X P), where SRA is spermidine specific radioactivity (cpm/pmol) and P is protein content (mg). The factor % is used because only one of the two labeled methylenes from spermidine is incorporated into hypusine (2, 24). The abbreviations used are: CH, cycloheximide; SPD, spermidine.

TABLE I1 Distribution of deox.vh,vpusine and hypusine
Cells were treated as described in the legend to Fig. 2, A-D. The quantity of newly formed deoxyh-ypusine and h-ypusine were estimated as described in the legend of Fig. 2 after ion exchange chromatographic separation of these amino acids. In parentheses are given the percentages of each of the two components in newly formed deoxyh?rpusine + hypusine.  (Table 11, part C), and -70% in cells initially depleted of and then restored with spermidine (Table  11, part D ) . T h e percentages of ["Hldeoxyhypusine decreased with incubation time and were lower than 5 5 after 24 h (not shown) in all cases.
In spite of the inherent technical limitations involved in the accurate determination of hypusine and deoxyhypusine from radioactivity measurements as described in the legend of Fig. 2, the estimated rates in untreated exponent.ially growing CHO cells are in fair agreement with those values (1-2 pmol/mg protein/h) previously reported for CHO cells and HTC cells (6,8). The rate of deoxyhypusine + hypusine synthesis declined only slightly after 24 h of DFMO treatment (compare A and R in Fig. 2 , O ) . Thus, it appears that there is no significant inhibition of either the translation of eIF-4D mRNA or the modification of eIF-4D precursor I (Scheme 1) up to this point. Marked depression of deoxyhypusine + hypusine formation was found to occur after 42 h at the time when cellular spermidine was depleted (compare A and C in Fig. 2, 0 ) . This reduction in modification step 1 (Scheme 1) is probably responsible for the decrease in protein-bound h-ypusine (Fig. 1) and the accumulation of eIF-4D precursor I ( Figs. 2 and 3 ) in DFMO-treated cells.
Inhibition of translation by cycloheximide caused the complete arrest of new synthesis of deoxyhypusine + hypusine in untreated CHO cells (Fig. 2 A , A), and also in cells pretreated with DFMO for 24 h (Fig. 2R, A). This effect by cycloheximide was seen in cells from the early to the late exponential stage of growth irrespective of cell densities (not shown). The failure of untreated cells and cells with reduced spermidine content (DFMO 24 h) to form hypusine in the presence of cycloheximide probably results from the lack of substrate protein eIF-4D precursor I (Scheme l), rather than from inhibition of the enzymes involved. Certainly, the enzymes involved in hypusine production are functional in the presence of cycloheximide (Fig. 2, C and D, A). Thus the indication of the lack of eIF-4D precursor I accumulation supports the notion that eIF-4D precursor I undergoes post-translational modificat.ion during or immediately following translation in control cells and even in cells with reduced spermidine content (DFMO, 24 h). Furthermore, the complete arrest of radiolabeling of eIF-4D by cycloheximide (Fig. 2, A and R, A) suggests that hypusine is produced only de nouo in eIF-4D precursor I and that there is no pathway for exchange or turnover of the 4amino-2-hydroxybutyl moiety in the hypusine residue of eIF-4D.
In contrast to control cells and cells wit.h reduced spermi-  Fig. 2C, A). Whereas the addition of exogenous spermidine (2.5 PM) did not cause changes in the rate of deoxyhypusine + hypusine synthesis in control cells (not shown), replenishment of this polyamine to the control level in spermidinedepleted cells (DFMO, 42 h) by addition of 2.5 P M unlabeled spermidine (8) together with ["Hlspermidine, caused a 5-to 10-fold elevation of the initial rate of synthesis of deoxyhypusine + hypusine (compare C and D in Fig. 2, 0 ) . T h e finding of no apparent inhibition by cycloheximide at 1.3 h (Fig. 2 0 , A) is probably due to the small amount of eIF-4D precursor I newly translated during this period compared to what may be a pre-existing pool of precursor 1. The partial inhibition seen at 5 h (Fig. 211, uertical broken arrow) is likely the result of arrest of eIF-4D precursor I translation by cycloheximide over the incubation period of 5 h. Certainly, the cycloheximide resistant labeling of deoxyhypusine + hy-pusine substantiates the existence of a pool of eIF-4D precursor I in the spermidine-depleted cells. Evidently, the accumulation of this precursor occurred when the rate of modification (step 1 in Scheme 1) dropped below that of the translation of eIF-4D mRNA as a result of reduction of cellular spermidine from -10% (DFMO, 24 h) to an undetectable level (DFMO, 42 h). The extent of eIF-4D precursor I accumulation during the period between 24 and 42 h of DFMO treatment may be estimated as approximately 9 pmol/mg of protein from the maximum value of cycloheximide-resistant synthesis of deoxyhypusine + hypusine (Fig. 2D, vertical solid arrow).
Additional evidence for accumulation of eIF-4D precursor I in spermidine-deprived cells was provided by direct comparison of labeling of [3H]deoxyhypusine + [3H]hypusine and ['HH]leucine in eIF-4D from untreated CHO cells with that from cells pretreated with DFMO for 42 h. Labeling was carried out for 2 h with 3H-labeled spermidine and 3H-labeled leucine (Fig. 3). Cold spermidine (2.5 p~) was added to cells pretreated with DFMO (42 h) together with the two radiolabeled compounds in order to maintain the specific radioactivity of the spermidine pool of these cells comparable to that of control cells. Both [3H]deoxyhypusine and [3H]hypusine were found in the spots indicated by the solid arrows in essentially the same ratios as given in Table I1 (Fig. 3 F ) . No radiolabeling of eIF-4D was observed when control cells were incubated with ["Hlspermidine and [3H]leucine in the presence of cycloheximide (not shown). Thus a rapid modification of a pool of eIF-4D precursor I accumulated during spermidine depletion must have occurred upon replenishment of this polyamine in the absence (Fig. 3 E ) or presence (Fig. 3F) of cycloheximide. The post-translational modification independent of new translation seen in E and F of Fig. 3 is consistent with the data given in Fig. 2 0 and strongly supports the presence of a pool of eIF-4D precursor I in spermidine-depleted cells.

DISCUSSION
The data presented in this paper provide strong evidence that the post-translational modification of eIF-4D precursor I (step 1 in Scheme 1) normally occurs concomitant with or immediately after translation of eIF-4D mRNA in control CHO cells and also in cells with approximately one-tenth of their normal spermidine content (DFMO, 24 h). A coupled translation/hypusine modification has been proposed by Duncan and Hershey (12) on the basis of the effects of translation inhibition in HeLa cells on the incorporation of radioactivity from ["Hlspermidine into eIF-4D. However, we have detected the accumulation of a small amount of deoxyhypusine-containing eIF-4D precursor I1 in control CHO cells and in cells with reduced spermidine content ( A and B, respectively, in Table 11). Thus it appears that although deoxyhypusine synthesis is coupled to translation, immediate hydroxylation may not occur following the initial step of modification. Furthermore, we have found conditions under which the translation and the post-translational modification step 1 (Scheme 1) can be uncoupled. Our data suggest that in cells which are virtually depleted of spermidine, deoxyhypusine synthesis is no longer coupled to translation, the eIF-4D precursor I that contains no deoxyhypusine or hypusine accumulates and, upon replenishment with spermidine, rapid modification of preformed eIF-4D precursor I can occur. It is obvious from the data in Fig. 2 that the rate of modification (step 1 in Scheme 1) is significantly reduced when spermidine is depleted. Whether cellular polyamines regulate the translation of eIF-4D mRNA and its gene expression is not known. However, it may be concluded that the accumulation of eIF-4D precursor I can result from the differential effects of spermidine depletion on the translation and the modification (step 1 in Scheme 1). In the cells containing a pool of eIF-4D precursor I (42 h preincubation with DFMO, C and D in Fig. 2) spermidine appears to be the major factor determining the rate of deoxyhypusine + hypusine formation (modification step I in Scheme 1). The spermidine concentration dependence of hypusine synthesis in rat hepatoma cells was previously reported by Gerner et al. (8). It is not clear at present whether the effect of cellular spermidine is exerted at the level of substrate concentration or at the level of activities of the enzymes involved or both. It is obvious from the data of Table I1 that both radiolabeled deoxyhypusine and hypusine are detected in the early times after addition of [3H]spermidine. Although the quantity of deoxyhypusine is small compared to the total protein-bound hypusine ( Table I1 and Fig. I), the levels of deoxyhypusine in newly modified protein are higher in the cells with the higher initial rates of modification (step I , Scheme 1). This increased accumulation of unhydroxylated amino acid may result from a relatively stable rate of hydroxylation irrespective of the large increase in initial synthesis of deoxyhypusine. Alternatively, deoxyhypusine hydroxylation may also be regulated by the level of cellular polyamines and in an independent manner. This possible regulation is currently being investigated.
In spite of the strong evidence for accumulation of eIF-4D precursor I in spermidine-deficient cells, we have thus far failed to detect eIF-4D precursor I protein after separation of cellular proteins by two-dimensional gel electrophoresis. This may be due to the small quantity of this precursor, -0.018% of the total cellular protein, and the limited resolution of our two-dimensional gels. Enrichment of this protein fraction and use of specific antibodies may provide the means for direct identification of this precursor protein.
The physiological significance of the decreases in the biosynthesis and content of protein-bound hypusine, and of the accumulation of eIF-4D precursor I in DFMO-treated cells is not known. Depletion of putrescine and spermidine by DFMO leads to depression of total cell protein synthesis in HTC cells (14) and in mitogen-activated lymphocytes (15) and causes the inhibition of growth in various mammalian cells (16)(17)(18)(19). However, the mechanism by which polyamines contribute to the regulation of cellular proliferation is not clearly understood. In cells treated with DFMO, or in mutant cells defective in polyamine synthesis, the marked inhibition of replication is delayed by 1 to 3 generations after the depletion of putrescine and spermidine (16,18,20,21). On this ground it was speculated that the inhibition of proliferation may not be a direct consequence of polyamine depletion, but rather that polyamines may participate in the biogenesis of some cellular component(s) that is required in the replication (18). Since neither the function of eIF-4D in eukaryotic protein synthesis (22) nor the role of hypusine in this factor has been elucidated, it is as yet premature to attribute inhibition of cellular proliferation to a decrease in hypusine. In a recent report, Duncan and Hershey (12) suggest that changes in eIF-4D hypusine modification or abundance are not correlated with translation repression in HeLa cells by serum depletion, heat shock, or hypertonic shock. Although normally, eIF-4D precursor is largely modified to contain hypusine, our results show that the modification (step 1, Scheme 1) is regulated in DFMOtreated cells where cellular spermidine is drastically reduced. In this system, the modification may play a critical role in the modulation of protein synthesis and cellular proliferation. The isolation of eIF-4D precursor I from spermidine-deficient cells would provide the first step towards an understanding of the role of hypusine. Possession of this unmodified protein is certainly essential for in vitro studies of the mechanism of biosynthesis of deoxyhypusine and hypusine and its regulation.