Arginine-specific regulation mediated by the Neurospora crassa arg-2 upstream open reading frame in a homologous, cell-free in vitro translation system.

Translational control mediated by an upstream open reading frame (uORF) in the 5'-leader of the Neurospora crassa arg-2 mRNA was reconstituted in a homologous, cell-free in vitro translation system. A cell-free N. crassa system was developed that required the presence of cap and poly(A) on RNA for maximal translation and that was amino acid-dependent. The 24-codon arg-2 uORF, when placed in the 5'-leader region of capped and adenylated synthetic luciferase RNAs, conferred Arg-specific negative regulation in this system. Improving the uORF translation initiation context decreased luciferase production and only slightly increased the magnitude of Arg-specific regulation. Mutation of uORF Asp codon 12 to Asn, which eliminates Arg-specific regulation in vivo, eliminated regulation in vitro. Elimination of the uORF translation initiation codon also eliminated Arg-specific regulation. Arg-specific regulation in vitro appeared to be reversible. Control of RNA stability did not appear to be a primary component of Arg-specific regulation in vitro. Comparison of the effects of adding Arg to in vitro translation reactions with adding compounds related to Arg indicated that Arg-specific translational regulation was specific for L-arginine.

Translational control mediated by an upstream open reading frame (uORF) in the 5-leader of the Neurospora crassa arg-2 mRNA was reconstituted in a homologous, cell-free in vitro translation system. A cell-free N. crassa system was developed that required the presence of cap and poly(A) on RNA for maximal translation and that was amino acid-dependent. The 24-codon arg-2 uORF, when placed in the 5-leader region of capped and adenylated synthetic luciferase RNAs, conferred Arg-specific negative regulation in this system. Improving the uORF translation initiation context decreased luciferase production and only slightly increased the magnitude of Arg-specific regulation. Mutation of uORF Asp codon 12 to Asn, which eliminates Arg-specific regulation in vivo, eliminated regulation in vitro. Elimination of the uORF translation initiation codon also eliminated Arg-specific regulation. Arg-specific regulation in vitro appeared to be reversible. Control of RNA stability did not appear to be a primary component of Arg-specific regulation in vitro. Comparison of the effects of adding Arg to in vitro translation reactions with adding compounds related to Arg indicated that Arg-specific translational regulation was specific for L-arginine.
Translational control is an important regulatory mechanism that often depends on sequences within the transcript being translated (1). A significant subset of eukaryotic mRNAs, including many encoding polypeptides involved in growth control and development, contain upstream open reading frames (uORFs) 1 in their 5Ј leader regions. In many cases, these uORFs influence the level of translation (2)(3)(4)(5). The mechanisms by which uORFs regulate translation in eukaryotes are mostly unknown.
In several instances, uORF-mediated translational control has been demonstrated to occur in response to specific environmental conditions. Expression of Saccharomyces cerevisiae GCN4 is regulated by limitation for many different amino acids through a mechanism involving the translation of multiple uORFs in the mRNA (4,6,7). The predicted peptide sequences encoded in the GCN4 uORFs are not important for control. Expression of Neurospora crassa arg-2, which specifies the small subunit of arginine-specific carbamoyl phosphate synthetase, is subject to unique, arginine (Arg)-specific translational regulation (8). In Arg-containing medium, the translation of mRNA containing the wild-type arg-2 uORF decreases, and the decrease in translation is associated with a decrease in the number of ribosomes associated with the mRNA. The sequence of an uORF specifying a 24-residue peptide is critical for Argspecific translational control; an Asp to Asn codon change at codon 12 of the arg-2 uORF (D12N) abrogates this control (9,10). The corresponding S. cerevisiae gene, CPA1, contains a similar uORF whose sequence is also important for Arg-specific regulation, which presumably also occurs at the level of translation (11,12).
The relatively common occurrence of uORFs in eukaryotic mRNAs (2,3,13) suggest that uORFs often have roles in modulating gene expression. As with the arg-2 and CPA1 uORFs, the sequences of the peptides specified by uORFs in the transcripts for mammalian ␤ 2 adrenergic receptor (14), human S-adenosyl methionine decarboxylase (15), cytomegalovirus gp48 (16), and maize Lc (17) RNAs appear to be important for uORF function. The sequences of uORF-encoded peptides in prokaryotes can also be important for translational control, and common mechanisms may be involved in uORF-mediated translational control in both kingdoms (5).
Cell-free translation systems have been invaluable for addressing many mechanisms of translational control (1). Here we describe an amino acid-dependent N. crassa cell-free translation system that reconstitutes cap, poly(A), and uORF effects on translation. Although there have been previous descriptions of programmable N. crassa cell-free translation systems (18 -21), translational control has not been investigated using such a system. To our knowledge, these data represent the first instance in which translational control in response to the availability of a single amino acid has been reconstituted in a eukaryotic cell-free translation system.
BglII-and XhoI-digested megaprimer PCR products were gel-purified and ligated to BglII-and XhoI-digested vectors pHLucS4 (see Fig.  8A) and pHLucϩNFS4 (see Fig. 8C). pHLucS4 (23), which was based on pHST7, was from N. I. T. Zanchin and J. E. G. McCarthy (National Biotechnology Research Center, Federal Republic of Germany); pHLucϩNFS4 was constructed by digestion of pHLucS4 with NcoI and XbaI, and replacement of the luciferase coding region with the NcoI-XbaI fragment of pSPLucϩNF (Promega), which contained the LucϩNF luciferase coding region. The sequences of plasmid constructs were confirmed by sequencing both strands of the template plasmids.
Preparation of Synthetic RNA Transcripts-Synthetic RNA for N. crassa cox-5 (cytochrome oxidase subunit V) was obtained from plasmid pSRCOX5 (24), which had been linearized with HindIII. Synthetic RNA for N. crassa arg-2 was obtained from plasmid pAH4, which had been linearized with EcoRI. pAH4 contained the PvuII-EcoRI fragment of arg-2 cDNA obtained from pARCG228 (25) subcloned into PvuII-and EcoRI-digested pSP72. Luciferase RNA was obtained from a variety of linearized plasmids. pSPLucϩNF was linearized with EcoRI; plasmids T3LUC and T3LUCpA (26) were linearized with BamHI; and plasmids pHLucS4, pHLucϩNFS4, and their derivatives (see Fig. 8) were linearized with Ppu10I. Synthetic RNAs from pHLucS4 and pHLucϩNFS4 should contain a poly(A) tail of 30 adenylate residues, as encoded by the template; in practice, the poly(A) tail length is expected to be longer because T7 RNA polymerase slips when synthesizing poly(A) tracts (27).
Synthetic RNAs were prepared by run-off transcription of linearized DNA templates. Reactions (50 l) to synthesize capped RNA contained: 2 g of linearized template, transcription buffer (40 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 2 mM spermidine, 10 mM NaCl), 10 mM dithiothreitol, 0.5 mM each of ATP, CTP, and UTP, 2 Ci of [␣-32 P]UTP, 0.05 mM GTP, 0.5 mM m 7 G(5Ј)ppp(5Ј)G, 50 units of RNasin, and 50 units of T7, T3, or SP6 RNA polymerase as appropriate. These synthesis conditions yield high levels of capped RNA (28) as borne out by analyses of the effects of exogenous cap on the translation of capped RNA compared with uncapped RNA (see Fig. 4). Water was substituted for cap when uncapped RNA was synthesized. For RNA that was used in chemical stability experiments, the concentration of unlabeled UTP was lowered to 0.05 mM, and 20 Ci of [␣-32 P]UTP was added. In all experiments, a common stock of reactants without DNA or enzyme was prepared and aliquoted to separate tubes to which DNA and enzyme were then added. This ensured that RNAs prepared in parallel would be radiolabeled with [␣-32 P]UTP to the same specific activity. Reactions (50 l) were incubated for 1 h at 37°C. After incubation, 150 l of water and 200 l of 5 M NH 4 OAc were added, and the reactions were placed on ice for 15 min to selectively precipitate the RNA. Precipitated RNA was recovered by centrifugation and dissolved in 200 l of water. RNAs were reprecipitated with KOAc and ethanol, washed with 70% ethanol, and dissolved in 200 l of water.
Yields of RNA, and verification that the purified RNA was free of unincorporated nucleotides were determined by polyethyleneimine chromatography and by trichloroacetic acid precipitation. Intactness of RNA and further proof of successful synthesis was accomplished by ethidium staining of RNA following electrophoresis in formaldehyde agarose gels. Equal amounts of different RNAs that were prepared in parallel were used in translation studies. These amounts were determined by considering the size of each RNA, the fraction of U residues within each RNA, and the amount of each RNA synthesized as measured by incorporation of radiolabeled nucleotide.
Preparation of Cell-free Extracts for Translation-Wild-type N. crassa strain 74A-OR23-1VA was obtained from D. Perkins, Stanford University. Neurospora conidia were collected with water and germinated in Vogel's minimal medium/1.5% sucrose at a concentration of 1 ϫ 10 7 conidia/ml (8). To prepare cell-free extracts that were not dependent on the addition of amino acids, 1-liter cultures were incubated at 34°C with orbital shaking (200 rpm) for 6.5 h, and then mycelia were harvested by vacuum filtration onto Whatman 541 filter paper. To prepare cell-free extracts that were dependent on addition of amino acids, 1-liter cultures were incubated at 32°C for 8 h, and mycelia were harvested by vacuum filtration, resuspended in 1 liter of fresh growth medium, incubated an additional 1 h, and then harvested again. Although we do not understand why this latter growth regimen enables the production of amino acid-dependent cell-free translation extracts, it does so reproducibly. Following harvesting, mycelial pads were rinsed with ice-cold buffer A (30 mM HEPES-KOH, pH 7.6; 100 mM KOAc; 3 mM MgOAc; 2 mM dithiothreitol (29)). In the cold room, mycelia (10 g of wet weight) were combined with 10 g of acid-washed 0.5-mm glass beads and ground with a mortar and pestle with the gradual addition of 10 ml of grinding buffer (buffer A with protease inhibitors (25 g/ml p-amidinophenylmethylsulfonyl fluoride (Calbiochem), 5 g/ml each of pepstatin A, antipain, chymostatin, and leupeptin (Sig-ma)). The homogenized mycelia were centrifuged in a polycarbonate centrifuge tube for 10 min at 31,000 ϫ g at 4°C in an SS34 rotor. The supernatant was carefully removed, avoiding both the pellet and the fatty upper layer; it was chromatographed on a 2.0 ϫ 20-cm Sephadex G-25 Superfine column that was pre-equilibrated with buffer A. Fractions containing the peak A 260 were pooled. Typical preparations yielded 6 -7 ml of extract with an A 260 /ml ϭ ϳ50. When extracts were treated with micrococcal nuclease, they were first adjusted to 1 mM CaCl 2 ; then micrococcal nuclease was added to 50 units/ml, and the extracts were incubated for 10 min at 21°C. EGTA was then added to a final concentration of 2.5 mM to inhibit the nuclease. Amino acid-dependent extracts were not treated with micrococcal nuclease. Extracts were aliquoted to Eppendorf tubes, frozen with liquid N 2 , and stored at Ϫ80°C. Standard reaction conditions were determined by analyses of the effects of different temperatures and different levels of K ϩ and Mg 2ϩ on translation. Temperatures of 25-30°C appeared optimal for cell-free translation. Varying the concentrations of Mg 2ϩ and K ϩ affected the translation of capped, polyadenylated luc RNA (capLUCpA RNA) and uncapped, unadenylated luc RNA (LUC RNA). Concentrations of these ions were chosen to yield the greatest relative translation of capLUCpA RNA compared with LUC RNA.

Cell-free Translation and Analyses of Translation Products-Unless
Translation of RNA in nuclease-treated reticulocyte lysates (Promega) was accomplished according to the supplier's directions. Translation of RNA in S. cerevisiae cell-free extracts was as described (29).
Quantitation of [ 35 S]methionine-labeled polypeptides was accomplished using a Molecular Dynamics PhosphorImager following SDSpolyacrylamide gel electrophoresis separation of the reaction products. The luciferase activity produced in N. crassa and reticulocyte translation reactions was measured by thawing translation reactions on ice, adding 5 l of the thawed reactions (diluted with luciferase reaction buffer if necessary) to 50 l of LUC assay reagent (Promega), and immediately measuring photon production in a Beckman LS6500 scintillation spectrometer. Luciferase activity in S. cerevisiae extracts was measured with a Turner TD-20e luminometer (29). All of the data presented represent averages of duplicate or triplicate reactions of one experiment; standard errors are indicated for experiments that examined Arg-specific regulation. All experiments were repeated multiple times with similar results.

RESULTS
Characterization of the N. crassa Cell-free Translation System-Analyses of [ 35 S]methionine-labeled products from N. crassa and reticulocyte systems programmed with synthetic, capped cox-5, arg-2 or luc RNAs or containing no exogenous RNA indicated that these systems were comparable in activity (Fig. 1). The observed sizes of the COX5, ARG2, and LUC translation products were similar in each system and consistent with the predicted masses of these polypeptides. The yields of radiolabeled polypeptides were similar in these systems; measurements of luciferase enzyme activity produced were also comparable when similar amounts of capped, adenylated luciferase RNA (capLUCpA RNA) were added to N. crassa and reticulocyte translation systems (see below).
A time-course experiment to measure production of luciferase (Fig. 2) from capLUCpA RNA revealed that within the first 10 min, there was little luciferase production; between 10 and 30 min, luciferase production was linear; after 30 min, luciferase production began to level off. [ 35 S]methionine-labeling experiments and analyses of the synthesis of full-length, radiolabeled luciferase polypeptide yielded comparable results (data not shown). Synthesis of the full-length [ 35 S]methionine-labeled COX5 polypeptide was observed earlier (6 min). Because COX5 polypeptide is smaller than LUC polypeptide, the lag in polypeptide synthesis was likely to be due in part to the time required for nascent polypeptide elongation (data not shown).
The effects of cap and poly(A) on the efficiency of translation were examined in nuclease-treated N. crassa extracts over a wide range of RNA concentrations (Fig. 3A). The level of luciferase production was linearly proportional to the level of RNA used to program translation in all cases over RNA concentrations varying by several orders of magnitude. Uncapped, unadenylated RNA (LUC RNA) was least efficiently translated. The addition of cap to RNA (capLUC RNA) increased translation. The addition of poly(A) alone to RNA (LUCpA RNA) stimulated translation more than addition of cap alone. The addition of both cap and poly(A) to mRNA (capLUCpA RNA) had synergistic stimulatory effects on translation, and capLUCpA RNA translated best. CapLUCpA typically translated more than 2 orders of magnitude more efficiently than LUC RNA.
Another observation consistent with the cap-and poly(A)-dependent translation that was seen in nuclease-treated extracts was that the addition of exogenous cap analog to N. crassa translation reactions strongly inhibited the translation of capLUC RNA (Fig. 4). The addition of poly(A) to RNA relieved inhibition of translation by cap analog; translation of LUCpA RNA was least affected by the addition of analog, and translation of capLUCpA RNA was less affected than capLUC RNA (Fig. 4).
Removal of endogenous RNA by pretreatment of N. crassa extracts with micrococcal nuclease was necessary for detecting [ 35 S]methionine-labeled translation products ( Fig. 1 and data not shown). However, this was not necessary for luciferase measurements. Thus, the addition of cap and poly(A) to RNA had similar, synergistic effects on translation of luciferase in N. crassa extracts that were not treated with nuclease (Fig. 3B) as extracts that were nuclease-treated (Fig. 3A). Commercial reticulocyte lysate did not show cap-and poly(A)-dependent effects on translation under the recommended translation conditions (Fig. 3B).
Direct measurement of the chemical stability of translated RNAs showed that RNA was degraded during incubation in N. crassa translation reactions (Fig. 5). Cap and poly(A) both stabilized the RNA; cap had a greater effect than poly(A).

Effects of Upstream Open Reading Frames on Translation-
The standard N. crassa translation system prepared from cells growing in minimal medium was not dependent on the addition of exogenous amino acids for translation (Fig. 6), and the addition of extra Arg did not affect translation of uORF-containing RNAs (data not shown). Therefore, as described under "Experimental Procedures," we used a different set of growth conditions that enabled the preparation of amino acid-dependent N. crassa translation extracts to test the effects of adding Arg. The effects of adding different concentrations of 20 amino acids on production of luciferase in amino acid-dependent and -independent N. crassa translation reactions are shown in Fig.  6. The amino acid-dependent translation system enabled testing of Arg-specific translational regulation in vitro.
Both amino acid-independent and amino acid-dependent cell-free translation systems were used to examine the effects of the arg-2 uORF on translation of the luciferase coding regions in RNA. Wild-type and mutant arg-2 sequences (Fig. 7) were placed upstream of the luciferase coding regions in either of two vectors (Fig. 8, A and C). One vector contained the wild-type luciferase coding region, LUC; the other contained a luciferase coding region modified by site-specific mutagenesis to alter codon usage and eliminate several restriction enzyme cleavage sites, LUCϩNF. Constructs were used to produce capped, polyadenylated RNA that contained the luciferase coding region and (i) no arg-2 sequence; (ii) the wild-type arg-2 5Ј leader containing the uORF (wild-type uORF); (iii) an arg-2 5Ј leader containing the wild-type arg-2 uORF in a better predicted initiation context (1 uORF); (iv) the wild-type arg-2 5Ј leader containing the arg-2 uORF with the Asp to Asn change that abrogated regulation in vivo (D12N); (v) a double mutant containing the D12N uORF in a better predicted initiation context (D12N 1 uORF); and (vi) a mutated arg-2 5Ј leader lacking the uORF AUG codon (⌬AUG).
Equal amounts of the capped, adenylated RNAs obtained from the vectors in Fig. 8A were compared in the amino acidindependent, micrococcal nuclease-treated N. crassa in vitro translation system and nuclease-treated S. cerevisiae and reticulocyte systems. Transcripts containing uORFs were translated less well than transcripts lacking uORFs in the N. crassa system, as also observed in the amino acid-dependent system (discussed below); similar results were observed with the S. cerevisiae translation system but not with the reticulocyte system, which did not appear to respond to uORFs (data not shown).
In amino acid-dependent translation reactions, the addition of Arg to a final concentration of 10 M and the addition of the other 19 amino acids to final concentrations of 10 M were sufficient for near maximal translation of luciferase (Fig. 6), enabling testing of the effects of adding 10 or 500 M Arg on the translation of uORF-containing and control RNAs (Fig. 8B). Excess Arg did not affect translation of RNA obtained from a construct that lacks all arg-2 sequences (T7LUC). Arg reduced translation of luciferase in RNA containing the wild-type uORF in its original translation initiation context or in an improved initiation context upstream of LUC (Fig. 8B). A reproducible, slight increase in Arg-specific regulation was seen with constructs containing the uORF in an improved initiation context. Improving the translation initiation context of the uORF also decreased production of luciferase from the downstream initiation codon. In contrast to the wild-type uORF constructs, the D12N uORF construct did not show Arg-specific regulation. Eliminating the uORF initiation codon eliminated regulation and increased expression to the level observed in the construct lacking arg-2 sequences (Fig. 8B).
The data shown in Fig. 8B were highly reproducible. Considering the T7LUC, wild-type uORF, and D12N uORF constructs alone, five different batches of RNA translated in nine independently derived Arg-dependent N. crassa extracts gave similar results. Three independent batches of RNA synthesized from all five constructs have been translated in three independently derived Arg-dependent translation extracts with similar results.
The effects of uORFs on Arg-regulation were confirmed and extended using a second set of luciferase constructs in which LUCϩNF (Promega) has been substituted for LUC (Fig. 8D). In N. crassa extracts, RNAs containing the LUCϩNF polypeptide coding region instead of LUC produced similar yields of luciferase activity. The wild-type uORF construct, the 1 uORF construct and the D12N uORF construct showed effects on translation similar to those observed in the original LUC constructs (Fig. 8, compare B and D). An additional construct, in which the D12N mutation was placed in an uORF that had an improved initiation context (D12N 1 uORF), showed reduced translation of luciferase compared with the D12N mutation in the original uORF initiation context but still showed no Argspecific regulation (Fig. 8D).
The primary mechanism of Arg-specific regulation mediated by the arg-2 uORF in the in vitro translation system does not appear to involve changes to the RNA. First, the chemical stability of each of the types of RNAs used in Fig. 8B was measured by determining the amount of trichloroacetic acidsoluble radioactivity released from RNA over the course of translation reactions. There were no discernible differences among RNAs; the addition of excess Arg did not appear to affect the stability of any of these RNAs (data not shown). Second, the negative effects of Arg on translation of the wild-type uORF RNA appeared to be reversible. Translation reactions containing excess Arg that were programmed with RNA containing the wild-type uORF translated RNA at a reduced rate compared FIG. 7. Sequence (GenBank accession number J05512) of the arg-2 5 region with introns removed (25). Triangles indicate the major mRNA 5Ј ends in vivo. The amino acid sequences of the uORF and the N terminus of ARG2 are given above the nucleotide sequence. Mutations that were analyzed in the in vitro translation system are indicated. These include the change to D12N, changes that improve the translation initiation context of the uORF (1 uORF), and a change that eliminates the uORF start codon (⌬AUG). The 5Ј and 3Ј boundaries of the arg-2 region that was amplified by PCR for testing in vitro are boxed. with extracts without excess Arg (Fig. 9). 14-fold dilution of such translation mixtures with additional complete translation mixture lacking Arg after translation was initiated for 20 min resulted in an increased rate of RNA translation (Fig. 9). This rate was comparable with that observed in translation reactions initiated without excess Arg that were monitored at a similar time and was substantially higher than the rates observed in reactions diluted with reaction mixture containing excess Arg (Fig. 9). Similar results were obtained when reactions were diluted at 15 or 30 min instead of 20 min after incubation (data not shown). These data indicate that (i) degradation or irreversible modification of uORF-containing RNA was unlikely to be responsible for its reduced translation in the presence of excess Arg and (ii) the effects of Arg on translation of uORF-containing RNA appeared reversible.
How specific is Arg-specific regulation? The effects of compounds that might be expected to act similarly to Arg were examined by comparing the addition of 500 M of Arg or 500 M of each of these compounds. The addition of 150 M Arg was sufficient to observe maximal regulatory effects (data not shown). The addition of canavanine, an Arg analog that can be incorporated into polypeptides and, when incorporated, prevents their functional activity, resulted in a loss of luciferase activity in all cases, and its activity was not evaluated. Arginine methyl ester and arginine ethyl ester, which can be hydrolyzed to Arg, conferred regulation, but the Arg biosynthetic precursors citrulline and ornithine did not, nor did the basic amino acids His and Lys. Homoarginine, which has a side chain that is one methyl group longer than Arg, did not confer regulation. The Arg-related compounds agmatine, L-argininamide, phospho-L-arginine, L-argininic acid, and D-arginine did not confer regulation. These data strongly indicate that Arg-specific translational regulation has a high specificity for sensing the level of L-arginine.

DISCUSSION
In vivo, Arg-specific regulation has effects on the expression of the N. crassa arg-2 and S. cerevisiae CPA1 genes specifying the small subunit of Arg-specific carbamoyl phosphate synthe-tase at both transcriptional and translational levels (8,25,30,31). Here we show that the N. crassa arg-2 uORF has a role in Arg-specific translational regulation using a homologous cellfree in vitro translation system programmed with synthetic RNA. The sequence of the uORF was critical for regulation in vitro, but the context of the uORF codon initiation codon was not. Furthermore, Arg-specific translational regulation appeared specific for L-arginine and was not elicited by related amino acids or by biosynthetic precursors to Arg. These data confirm that the arg-2 uORF modulates gene expression at the level of translation and to our knowledge represent the first demonstration of translational control in a eukaryotic in vitro system in response to the availability of a single amino acid.
Translation of RNA in the N. crassa cell-free translation system also was dependent on important features of eukaryotic RNA, cap, and poly(A). The addition of poly(A) to RNA stimulated translation in the N. crassa cell free system independently of the addition of cap to RNA, whereas the addition of both to RNA synergistically stimulated translation (Fig. 3). Although synergistic interactions between cap and poly(A) have been demonstrated in electroporated mammalian and fungal cells (32), only recently have such interactions been observed in vitro and only in S. cerevisiae cell-free translation systems (26,29). The effects of adding cap and poly(A) to RNA used to program N. crassa and S. cerevisiae translation reactions are similar, and, in both systems, the addition of poly(A) to RNA relieves the inhibitory effects of adding exogenous cap analog to translation reactions. In the S. cerevisiae system, these results have been combined with additional studies to support the interpretation that cap and poly(A) stimulate translation initiation by recruiting different RNA-binding proteins (29), and it is likely that a similar situation holds in N. crassa.
In the N. crassa cell-free system, the presence of cap and poly(A) on mRNA appear to affect its chemical stability ( Fig. 5 and data not shown). In S. cerevisiae, similar effects have been reported in some studies (33) but not others (26,29). The reasons why cap and poly(A) have effects on RNA stability in some in vitro studies but not others remains unclear, but a variety of studies have implicated that these features of the RNA molecule are important in modulating RNA stability (34,35).
The in vitro data indicate that the arg-2 uORF has sequenceindependent and sequence-dependent effects on translation. Thus, either the wild-type uORF or the D12N uORFs in the wild-type translation initiation context reduce translation of the downstream luciferase coding region (Fig. 8, B and D). The uORF's wild-type initiation context is not typical for N. crassa (8); changing the initiation context of either uORF to one resembling preferred N. crassa initiation contexts (36) further reduces translation of the downstream luciferase coding region. These data are consistent with the scanning model for translation initiation (37,38).
Arg-specific and uORF-sequence-dependent regulation is observed in vitro in addition to sequence-independent uORF effects. Arg-specific regulation in wild-type cells mediated through the arg-2 uORF is approximately 2.5-fold in vivo, based on measurements of accumulated polypeptide products (8 -10); a similar level of regulation is observed in vitro, based on measurements of accumulated luciferase product (Fig. 8). Measurements of relative rates of ARG2 polypeptide synthesis in vivo immediately after switching cells from minimal medium to fresh minimal medium or to Arg-containing medium indicate a 2.5-fold reduction in the rate of polypeptide synthesis upon exposure to Arg (8); a similar reduction in the rate of luciferase synthesis from uORF-containing RNA is observed in vitro when translation reactions contain excess Arg (Fig. 9), based on comparing the rates of synthesis of reactions containing 10 M or 150 M Arg.
Our combined data show that Arg-specific translational regulation can be reconstituted in an amino acid-dependent N. crassa cell-free translation system. Our findings show that the level of regulation observed in vitro is very similar to that observed in vivo and that regulation is highly specific for Larginine. The mechanism of Arg-specific translational control remains to be elucidated; the amino acid-dependent in vitro system, in which translational effects can be monitored independently of transcriptional effects, should provide an invaluable tool for combining biochemical and genetic approaches to determining the details of this mechanism.