The upstream open reading frame of the mRNA encoding S-adenosylmethionine decarboxylase is a polyamine-responsive translational control element.

S-Adenosylmethionine decarboxylase (AdoMetDC) is a key enzyme in the pathway of polyamine biosynthesis. The cellular levels of the polyamines specifically regulate AdoMetDC translation through the 5′-leader of the mRNA, which contains a small upstream open reading frame (uORF) 14 nucleotides from the cap. Mutating the initiation codon of the uORF, which encodes a peptide product with the sequence MAGDIS, abolished regulation. In addition, the uORF is sufficient, by itself, to provide polyamine regulation when inserted into the 5′-leader of the human growth hormone mRNA. Changing the amino acid sequence at the carboxyl terminus of the peptide product of the uORF abolished polyamine regulation. In contrast, altering the nucleotide sequence of the uORF at degenerate positions, without changing the amino acid sequence of the peptide, did not affect regulation. Extending the distance between cap and uORF, thereby changing the rate of initiation at the initiator AUG of the uORF, did not alter polyamine regulation. When the uORF was extended so as to overlap, out of frame, the downstream major cistron, polyamine regulation was abolished. We propose that polyamines do not modulate the rate of recognition of the uORF but rather regulate interaction of the peptide product of the uORF with its target.

Polyamines (putrescine, spermidine, and spermine) are a group of low molecular weight compounds necessary for the optimal growth of all cells, eukaryotic and prokaryotic (for reviews, see Marton and Morris (1987) and Jä nne et al. (1991)). Depletion of cellular polyamine levels has been shown to lead to decreased cell growth and alterations in cell differentiation (Pegg and McCann, 1982;Pegg, 1988;Marton and Pegg, 1995). On the other hand, excessive levels of polyamines may have toxic effects (reviewed in Heby and Persson (1990); Morris (1991)). Deregulation of polyamine synthesis has been shown to cause neoplastic transformation of cultured cells and a propensity toward tumor development in transgenic animals (Auvinen et al., 1992;Moshier et al., 1993;Megosh et al., 1995). Probably because of the ramifications of deregulation, polyamine levels are tightly regulated by a variety of mechanisms, including feedback regulation of expres-sion, activity, and stability of key enzymes as well as polyamine degradation and excretion from cells (reviewed in Heby and Persson (1990); Morris (1991); Large (1992); Grillo and Colombatto (1994)).
There are two key regulated enzymes in the pathway of polyamine biosynthesis, ornithine decarboxylase and S-adenosylmethionine decarboxylase (AdoMetDC). 1 The levels of these enzymes are regulated not only by exogenous signals (such as growth factors, hormones, and tumor promoters), but also by the endogenous level of the polyamines. AdoMetDC catalyzes the formation of an intermediate necessary for the conversion of putrescine to spermidine and spermine and is located at an important branch point in the metabolism of S-adenosylmethionine. The intracellular levels of polyamines influence AdoMetDC expression at multiple steps, including mRNA level, translation, and protein half-life Pegg, 1985, 1986;Pajunen et al., 1988;White et al., 1990), forming feedback loops to maintain the normal concentration of polyamines in cells.
The translation of AdoMetDC mRNA in reticulocyte lysates has been reported to be more strongly inhibited by increasing concentrations of polyamines than general protein synthesis (Kameji and Pegg, 1987). In intact cells, several studies have focused on translational control of AdoMetDC by polyamines, using polyamine-depleting drugs, such as ␣-difluoromethylornithine (DFMO), a potent irreversible inhibitor of the first enzyme in polyamine biosynthesis, ornithine decarboxylase (for review, see Marton and Pegg (1995)). Intracellular polyamine levels influence the translation of AdoMetDC mRNA in vivo, and the major effect seems to be at the initiation step (White et al., 1990). More recently, it has been demonstrated that the 5Ј-leader of AdoMetDC mRNA mediates this polyamine regulation, pointing to the presence of a polyamineresponsive element in this region of the mRNA .
The AdoMetDC 5Ј-leader is unusually long (330 nucleotides) with a small upstream open reading frame (uORF) located 14 nucleotides from the cap (Hill and Morris, 1992;Pulkka et al., 1993;Waris et al., 1993). This uORF plays a dominant role in repressing AdoMetDC translation in T cells (Hill and Morris, 1992) through a mechanism that is critically dependent on the unique peptide sequence (MAGDIS) coded by the uORF. Missense mutations in the 3Ј-terminal three codons of the uORF relieve translational suppression, while synonymous changes, having no effect on the coding status, retain suppressive activ-* This work was supported by National Institutes of Health Grants CA39053 (to D. R. M.) and CA18138 (to A. E. P.). 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.
¶ To whom correspondence and reprint requests should be addressed. ity (Hill and Morris, 1993). The uORF is largely ignored in nonlymphoid cells because of its proximity to the 5Ј-end of the mRNA, resulting in efficient translation of the downstream cistron. In T cells, the uORF is efficiently recognized and translated, resulting in strong suppression of downstream translation. This cell-specific translational control involves regulation of recognition and initiation at the uORF but not modulation of the interaction of the peptide product of the uORF with its suppressive target (Ruan et al., 1994) In the present study, we identify the uORF sequence as the major polyamine-responsive element in the 5Ј-leader of AdoMetDC mRNA. The initiation codon and the specific peptide coding sequence are required to mediate the polyamine response. Polyamine regulation of AdoMetDC translation seems to be mediated through modulating the interaction of the inhibitory peptide product of the uORF with its target and is independent of the position of the uORF relative to the mRNA 5Ј-cap. This mode of regulation is notably different from that utilized in cell-specific translational control.

EXPERIMENTAL PROCEDURES
Chimeric Constructs-Chimeras between AdoMetDC and human growth hormone (hGH) were described in previous publications Morris, 1992, 1993;Ruan et al., 1994). Briefly, the various modified 5Ј-leaders of AdoMetDC were synthesized by polymerase chain reactions (Saiki et al., 1985) using pRS327 (Hill and Morris, 1992) as template. We inserted the modified 5Ј-leaders (with BamHI sites added on each end) into the BamHI site of pRhGH⌬5ЈTL (Hill and Morris, 1992) by standard methods (Sambrook et al., 1990). The sequence of each construct was verified from the middle of the promoter through the hGH translational initiation site using the dideoxy chain termination method (Sanger et al., 1977) with a sequencing kit from U.S. Biochemical Corp.
Cell Culture-Chinese hamster ovary (CHO) cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 200 M L-proline, 100 units/ml penicillin and 50 g/ml streptomycin. In experiments in which polyamines were depleted, a final concentration of 5 mM DFMO was added in the media.
Polyamine Levels-CHO cells were seeded at 50,000 cells/plate in 35-mm plates in the presence or absence of 5 mM DFMO. For some samples, 10 M spermidine and 1 mM aminoguanidine (to inhibit the action of oxidase present in serum) were added 24 h later. After 72 h in culture, the cells were washed with ice-cold 1 ϫ Tris-buffered saline (25 mM Tris-HCl, pH 7.4, 137 mM NaCl, 5 mM KCl, 0.7 mM CaCl 2 , 0.5 mM MgCl 2 , and 0.6 mM Na 2 HPO 4 ) twice, scraped into 1.5-ml tubes, and harvested by centrifugation. Cell pellets were extracted with 10% (w/v) trichloroacetic acid. Polyamine content was determined as described previously (Pegg et al., 1989).
[ 35 S]Methionine Pulse Labeling of CHO Cells-Cells were cultured as described for polyamine determination. Triplicate cultures at each time point were washed twice with culture medium lacking methionine, and medium containing [ 35 S]methionine was added (600 l of medium containing 10% fetal bovine serum with 2 Ci of [ 35 S]Met per 35-mm plate). 5 mM DFMO was added when indicated. After incubation at 37°C for 90 min, cells were washed twice with ice-cold 1 ϫ TBS, precipitated with 10% trichloroacetic acid, and scraped into 1.5-ml tubes. After harvesting by centrifugation, the cell pellet was washed with 10% trichloroacetic acid twice, once with an ethanol:ether mixture (1:1), and once with ether. The dry pellet was dissolved in 3 mM KOH, and the protein concentration of the cell pellet was measured using the standard Bradford method (Bio-Rad). The radioactivity in each sample was counted using a liquid scintillation counter.
DNA Transfection-CHO cells were transfected using lipofectamine (Life Technologies, Inc.) according to the manufacturer's instructions. Approximately 120,000 cells were seeded per 35-mm plate and allowed to grow for 24 h. For cells depleted of polyamines, a final concentration of 5 mM DFMO was added at plating and was present throughout the transfection. Cells were transfected with 5 g of the indicated construct in 1 ml of serum-free medium containing 6 l of lipofectamine. After incubation for 5 h at 37°C, the solution was replaced with 2 ml of fresh culture medium, and the cultures were incubated for 48 h before harvesting for analysis. In experiments where polyamines were restored to depleted cells, a final concentration of 10 M spermidine (with 1 mM aminoguanidine) was added at the time of transfection.
Polysome Analysis-For analysis of polysomes, we isolated and pu-rified RNA from polysomes fractionated by sucrose gradient centrifugation as described previously (Mach et al., 1986). Cytosolic extracts from CHO cells were prepared and layered onto the tops of 0.5-1.5 M sucrose gradients. After centrifugation at 36,000 rpm in a Beckman SW-40 rotor for 110 min, we fractionated the gradients into 12 1-ml fractions using an ISCO Density Gradient Fractionator (model 185), while monitoring absorbance at 254 nM. We purified RNA from 400-l aliquots of the 12 fractions and used them to prepare Northern blots for analysis. To analyze for chimeric hGH mRNA, we hybridized the blots with the 300-base pair SmaI/SacI fragment from the pRNN construct (Hill and Morris, 1992). To analyze for endogenous AdoMetDC mRNA, we hybridized the blots with the 400-base pair EcoRI/HindIII fragment from construct pSDC.4 -34 (Hill and Morris, 1992). To analyze for endogenous actin mRNA, we hybridized the blots with the 1.3-kilobase pair PstI fragment from construct pBA1 (Hill and Morris, 1992). All of the DNA probes were radiolabeled with [ 32 P]dCTP by random primed labeling (Feinberg and Vogelstein, 1984).
Rates of Accumulation of hGH Protein-Samples of medium were removed from cultures at 48 h after transfection. hGH accumulation in the media is linear with respect to time over this time period (Hill and Morris, 1992). Aliquots were diluted with fetal bovine serum and assayed for hGH using the double antibody binding assay using Allegro kit (Nichols Institute, San Juan Capistrano, CA) (Selden et al., 1985) according to the manufacturer's instructions. In order to normalize the amount of hGH produced to the amount of hGH mRNA in the cells (Hill and Morris, 1992), total RNA from the transfected cultures was used to prepare Northern blots, which were probed for chimeric hGH mRNA.
RNA Analysis by Northern Blots-Total RNA was collected from transfected cultures following the protocol of Chomczynski and Sacchi (1987). Cells were directly lysed on plates using 200 l of denaturing solution (4 M guanidine thiocyanate, 0.5% Sarkosyl, 25 mM sodium citrate, pH 7.0), the plates were washed with 200 l of denaturing solution, and the wash was combined with the initial extraction solution. To monitor RNA recovery during the subsequent mRNA purification steps, a truncated form of hGH mRNA (a 300-base pair SacI/SmaI fragment of hGH coding region) was synthesized in vitro, and equal amounts were added to the initial cell lysates. Samples were extracted with phenol and chloroform. After precipitation with ethanol, pellets were washed twice with 70% ethanol, dissolved in RNase-free H 2 O, and digested with DNase I for 10 min at 37°C (Hill and Morris, 1992). RNA loading buffer was added, and the RNA samples were subjected to electrophoresis on 1.2% agarose gel in the presence of formaldehyde. The Northern blots were hybridized to a hGH probe radiolabeled with [ 32 P]dCTP by random primed labeling (Feinberg and Vogelstein, 1984). The hGH probe was derived from a 300-base pair SacI/SmaI fragment of the hGH gene from pRNN plasmid (Hill and Morris, 1992). The chimeric hGH mRNA and the truncated hGH mRNA signals were quantitated by PhosphorImager scanning (model SF, Molecular Dynamics, Sunnyvale, CA), and the amount of hGH chimeric mRNA was corrected for recovery by the level of truncated hGH mRNA. The normalized mRNA levels were used to normalize the amount of hGH protein produced to the cellular content of hGH mRNA.

Influence of DFMO Treatment on Cellular Polyamines and
Protein Synthesis-DFMO, which inhibits ornithine decarboxylase activity (Pegg, 1986), has been shown previously to deplete polyamine levels and decrease the cellular rate of protein synthesis (Pegg and McCann, 1982;Pegg, 1986Pegg, , 1988. Treatment of CHO cells with DFMO for 72 h results in depletion of polyamines, especially spermidine (Table I). The cellular levels of spermidine and spermine could be maintained in cells

Effect of DFMO on polyamines in CHO cells
Cells were grown for 24 h as described under "Experimental Procedures." The medium was then changed, and the cells were allowed to grow for 48 h, when they were harvested and assayed for polyamine content. DFMO was added at the time the cells were plated, and spermidine was added when medium was changed.
At various times after initiating treatment of CHO cells with DFMO, the rate of protein synthesis was assessed by labeling with [ 35 S]methionine (Fig. 1A). There was significant inhibition of methionine incorporation at 24 and 48 h after the start of DFMO treatment, and this inhibition reached approximately 10-fold by 72 h. This inhibition of protein synthesis by DFMO treatment is also reflected in reduced accumulation of cellular protein in the cultures (Fig. 1B).
Influence of Polyamine Level on AdoMetDC Translation-Cytosolic extracts were prepared from DFMO-treated and control CHO cells to analyze the distribution of AdoMetDC mRNA in polysomes. The extracts were fractionated by centrifugation in sucrose gradients, and AdoMetDC mRNA was analyzed on Northern blots (Fig. 2). In untreated cells, AdoMetDC mRNA is broadly distributed from monosomes to polysomes of small size. When cells were treated with DFMO, AdoMetDC mRNA distribution is more narrow and centered in the region of five ribosomes per molecule. The increase in size of polysomes containing AdoMetDC mRNA in polyamine-depleted cells suggests that polyamines regulate the number of ribosomes associated with AdoMetDC mRNA and hence regulate the relative efficiency of translation of this mRNA. Consistent with previous results (White et al., 1990), there is no influence of polyamine depletion on the distribution of actin mRNA in polysomes (Fig.  2) or on the absorbance profile of the sucrose gradients at 254 nm (not shown).
Previous results  demonstrated that translational control of AdoMetDC was mediated through the 5Ј-leader of its mRNA. As shown in Fig. 2, distribution of chimeric mRNA containing the 5Ј-leader of AdoMetDC mRNA (from construct pRS327) mimics that of endogenous AdoMetDC mRNA in both untreated and DFMO-treated CHO cells. This chimeric mRNA moved from small and medium polysomes onto large polysomes after 72 h of DFMO treatment. By comparison, a control mRNA containing hGH 5Ј-leader (from construct pRNN) remained essentially at the same position in polysomes after cells were treated with DFMO, similar to the behavior of endogenous actin mRNA. Hence, the AdoMetDC 5Ј-leader governs the regulation by polyamines of ribosome loading onto the AdoMetDC mRNA.
Cells were transfected with constructs pRS327 and pRNN under different conditions of polyamine depletion (see Table II). We measured the concentrations of hGH in the culture media 48 h after transfection (Hill and Morris, 1992). The hGH concentrations were then normalized to the levels of mRNAs from the respective constructs to obtain a measure of relative translation. As shown in Table II, depletion of polyamines causes 86% decrease (ratio of DFMO to none is 0.14) in the translation of the mRNA containing the hGH 5Ј-leader. This was expected, based on the reduction in overall cellular protein synthesis demonstrated above. However, when hGH mRNA is instead downstream of the AdoMetDC 5Ј-leader, expression was reduced by only 19% (ratio is 0.81). Therefore, the AdoMetDC/ hGH chimera is roughly 6-fold more resistant to translational inhibition caused by polyamine depletion than is general protein synthesis represented by construct pRNN. From experiment to experiment the ratio (DFMO/non) was found to vary from 0.5 to 0.8, which perhaps reflects a variation in the degree

FIG. 2. Distribution of mRNAs from endogenous AdoMetDC and transfected constructs in polysomes from untreated and DFMO-treated cells.
Cytoplasmic extracts were prepared and centrifuged through 0.5-1.5 M sucrose gradients for 110 min. The gradients were divided into 12 fractions from top (1) to bottom (12), and the distribution of the mRNAs was analyzed by Northern blots (see "Experimental Procedures"). The positions of the monosome (80S) and pentasome (5-mer) peaks were determined from the absorbance profiles of the sucrose gradients at 254 nm (data not shown). of polyamine depletion. However, in all instances, the AdoMetDC 5Ј-leader was 4 -6-fold more resistant to DFMO treatment than the hGH 5Ј-leader. The addition of spermidine to the cultures prevented the DFMO-induced decrease of translation efficiency for both constructs (Table II), demonstrating that the effect on translation is specifically caused by polyamine depletion.
Sequence Elements in the AdoMetDC 5Ј-Leader Involved in Polyamine Regulation-Constructs containing mutations in several potentially important areas of AdoMetDC 5Ј-leader were tested. For experimental purposes, we divided the 5Јleader into three regions; region A is the 14 nucleotides between 5Ј-cap and the uORF, region B is the uORF itself, and region C includes the long intercistronic region between the uORF and the downstream cistron (hGH in these constructs). In construct pRS362 (Fig. 3), region A was replaced with 47 nucleotides from hGH 5Ј-leader. When this construct was transfected into CHO cells, polyamine depletion only slightly reduces its expression (12% reduction), while expression of control construct pRNN (hGH leader) showed a 90% reduction, showing that altering region A does not abolish polyamine regulation. Next, we tested constructs containing large deletions in region C. Internal deletions in constructs pRS103 and pRS66 collectively covered the entire region C. As shown in Fig.  3, expression of constructs pRS103 and pRS66 after polyamine depletion were, respectively, 64 and 43% of those under normal culture conditions; in this experiment, those values were comparable with that with wild type AdoMetDC 5Ј-leader. Therefore, we conclude that sequences in region C are not involved in polyamine regulation.
To test the uORF region, the initiator AUG for the uORF was mutated into GUG in construct pRS(G 316 )327, thereby eliminating the uORF. When this construct was transfected into CHO cells, its expression dropped by 86% upon polyamine depletion, similar to the result with pRNN. To further test the involvement of the uORF in polyamine regulation, we transferred the uORF into the hGH 5Ј-leader, 14 nucleotides from the cap, generating construct pR(SuORF)hGH. This construct was much more resistant to polyamine depletion, which reduced expression by only 13%, as compared with 89% for the parent pRNN construct. These results suggest that the uORF is the single important element in the AdoMetDC 5Ј-leader that mediates polyamine regulation and that the uORF needs to be translated in order to function as the mediator of polyamine regulation.
Influence of the uORF on Translational Regulation by Polyamines-The coding capacity of the three 3Ј-terminal codons of the AdoMetDC uORF is critical for suppressing translation (Hill and Morris, 1993). To test the role of the coding sequence of uORF in mediating polyamine regulation, we altered these three codons (Table III, top) and tested the constructs in CHO cells. Changing the fourth codon, which encodes aspartic acid, to arginine abolished polyamine regulation (see construct pRS(D/R)327 in Table III (middle)). Similarly, alteration of the fifth codon from isoleucine to alanine or the sixth codon from serine to alanine interfered with polyamine regulation (constructs pRS(I/A)327 and pRS(S/A)327 in Table III (middle)).
To test whether the nucleotide sequence or the coding capacity of the last three codons is required to be conserved for polyamine regulation, we changed six different nucleotides within the uORF while retaining the amino acid sequence of the putative peptide (construct pRS(WuORF)327 in Table III (top)). The translation efficiency of this construct is at least as high as the wild type construct after polyamine depletion (Table III, bottom). These results suggest that the regulation mediated by polyamines requires preservation of the amino acid sequence of the putative product but not the nucleotide sequence of the uORF itself.
We have previously shown there is no influence on translation by extending the uORF of the AdoMetDC leader so that it overlaps, out of frame, the downstream cistron, regardless of the cell type tested (Ruan et al., 1994). This construct (pRSTTG, Fig. 4) was tested for polyamine regulation and compared with the wild type pRS327 as well as pRS66. As Fig.  4 shows, extending uORF into the downstream cistron had little or no effect on expression in the absence of DFMO but abolished polyamine regulation. This result clearly distinguishes polyamine regulation from the cell-specific regulation described previously (Ruan et al., 1994). The result is also consistent with the conclusion that the sequence of the peptide product from uORF needs to be preserved to achieve efficient polyamine regulation.

DISCUSSION
Polyamines are required for normal cell growth and protein synthesis (Löwkvist et al., 1987;Bitonti et al., 1988;Mihm et al., 1989). In the present study, treatment of CHO cells with the inhibitor of ornithine decarboxylase, DFMO, resulted in polyamine depletion and caused a dramatic reduction in the rate of protein synthesis. Against this background of general inhibition of protein synthesis, the translation of AdoMetDC mRNA is sustained (Persson et al., 1989;Kameji and Pegg, 1987;White et al., 1990;Stjernborg et al., 1993). Here, we define in detail those aspects of mRNA structure that are required for translational control of AdoMetDC by polyamines.
The maintenance of the rate of AdoMetDC translation during polyamine deprivation comes about through a specific elevation of the rate of translational initiation on this mRNA relative to the rate of polypeptide chain elongation (White et al., 1990). This regulation is mediated, at least in part, through the 5Ј-leader of the mRNA . These conclusions are supported in the present study through the use of chimeric constructs containing hGH as a reporter gene. Translation from the intact wild type hGH mRNA was inhibited by DFMO treatment to an extent similar to total protein synthesis, while the chimera containing the AdoMetDC 5Ј-leader was more resistant to polyamine depletion. In agreement with the increase in relative translatability of the AdoMetDC/hGH chimera, its mRNA moved into larger polysomes upon polyamine depletion, paralleling the behavior of endogenous AdoMetDC mRNA (White et al., 1990). In contrast, the efficiencies of ribosome loading onto the mRNAs from the wild type hGH construct and the endogenous actin mRNA were unchanged.
The results from the current studies clearly show that the structural features necessary for polyamine regulation of AdoMetDC translation reside solely in the uORF of the 5Јleader. The clearest demonstration of this comes from the construct where polyamine regulation was conferred on the unreg-ulated hGH leader through insertion of the AdoMetDC uORF with no additional sequence. Previous work has shown that the AdoMetDC uORF suppresses translation of downstream cistrons through a mechanism that depends not only on its own translation but also on the sequence of the encoded peptide (Hill and Morris, 1993). These properties place the AdoMetDC uORF among the "sequence-specific uORFs," which have been proposed to suppress translation by a "ribosome-stalling" mechanism (Hill and Morris, 1993;Geballe and Morris, 1994;Cao and Geballe, 1996). In this mechanism, the nascent peptide encoded by the uORF interacts with a component of the translation machinery, and this interaction causes the ribosome to be stalled at the termination codon of the uORF. The stalled ribosome serves as a blockade, which prevents addi- FIG. 3. Influence on polyamine regulation of various mutations and deletions in AdoMetDC 5-leader. Construct pRNN contains the entire hGH 5Ј-leader, and construct pRS327 contains the entire wild type AdoMetDC 5Ј-leader (Hill and Morris, 1992). Details of the other constructs are described under "Results." Relative translation is defined as the rate of hGH protein accumulation divided by the level of the chimeric hGH mRNA. To normalize the data, the value for construct pRNN in untreated CHO cells was arbitrarily designated as 100. Striped boxes, BamHI sites; checkered box, AdoMetDC uORF; thick black line, AdoMetDC 5Ј-leader; thin black line, hGH 5Ј-leader.

TABLE III
Influence on polyamine regulation of mutations within AdoMetDC uORF At the top are shown mutations made in the nucleotide sequence of the uORF and the amino acid sequence of the peptide product. Relative translation is defined as the rate of hGH protein accumulation divided by the level of the chimeric hGH mRNA. The value for construct pRNN in untreated CHO cells was arbitrarily designated as 100.  tional ribosomes from loading onto the message, thereby suppressing downstream translation. Polyamine regulation was abolished by missense substitutions in the carboxyl half of the uORF sequence but not by synonymous mutations that extensively altered the nucleotide sequence but not the coding capacity of the uORF. These results demonstrate clearly that the specific peptide sequence encoded by the uORF, and not the nucleotide sequence, mediates polyamine regulation. The other important aspect of uORF-mediated polyamine regulation is that altering the distance between 5Ј-cap and uORF from 14 to 47 nucleotides did not change regulation by polyamines. This suggests that changing the efficiency of recognition of the uORF by the scanning ribosomes has little or no effect on the response to polyamine depletion. This is in clear contrast to the previously reported cell-specific regulation of AdoMetDC translation, which is mediated through recognition of the uORF and sensitive to the distance from the 5Ј-cap (Ruan et al., 1994).
The requirement in polyamine regulation for both translation of the AdoMetDC uORF and a unique peptide sequence is identical to the reported requirement for translational suppression by this uORF in T cells (Hill and Morris, 1993). As noted above, since polyamine control is independent of the distance of the initiator codon of the uORF from the cap, regulation by polyamines does not seem to be mediated by altering the efficiency of recognition of the uORF. To account for these results, we suggest that physiological levels of the polyamines stabilize specific interactions between the peptide encoded by the uORF and its target. According to this model, the strength of these interactions would be weakened in the polyamine-depleted state, leading to a reduction in the amount of ribosome stalling. By this mechanism, the inhibitory effect of the uORF would be relieved by polyamine depletion, with a consequent increase in downstream translation. When the peptide sequence encoded by the uORF is changed, the altered peptide no longer interacts with its target, and therefore, polyamines would no longer specifically influence downstream translation of the associated mRNA.
According to the above model, two factors contribute to regulation of ribosome stalling on the AdoMetDC uORF: the sequence of the peptide encoded by the uORF and the cellular level of polyamines. Changing either of these two factors would abolish both ribosome stalling and uORF-mediated regulation. One can envision three ways in which polyamines could stabilize the interaction between the nascent peptide and its target. First, polyamines might participate directly in this interaction, forming a stable ternary complex with peptide and target. Second, appropriate polyamine levels might provide a critical intracellular pH value required for this interaction. This possibility is suggested by the recent finding that polyamines may exert their physiological function through maintaining intra-cellular pH (Poulin and Pegg, 1995). Last, the interaction could be stabilized by a trans-acting factor, whose function is influenced in turn by polyamines. Regardless of which, if any, of these hypotheses is correct, the implication is that the specific structure of the nascent peptide encoded by the AdoMetDC uORF plays a critical role in polyamine-mediated regulation. The fact that the peptide sequence encoded by the AdoMetDC uORF (MAGDIS) is identical among the reported mammalian cDNAs is consistent with this model. Also consistent with a unique role for the MAGDIS peptide in regulation by polyamines is the observation that, in the currently available compendium of uORFs (Kozak, 1987), there is none that encodes this or a related sequence.
There are only a limited number of uORFs for which the mechanism of regulation is understood. There are several examples of uORFs that suppress downstream translation constitutively, and their recognition is regulated (for reviews, see Geballe and Morris (1994); Morris (1995)). The present example is of particular interest, since it appears that the suppressive influence of the uORF is regulated independently of the efficiency of its recognition. Another possible example where regulation by a uORF might be mediated through interactions with its peptide product is the CPA1 gene of yeast (Werner et al., 1987;Delbecq et al., 1994), in which feedback regulation by arginine is mediated at the translational level through a sequence-specific uORF. These two examples, AdoMetDC and CPA1, provide prototypes for regulated interactions between uORF-encoded peptides and their cellular targets.