A Highly Conserved Mechanism of Regulated Ribosome Stalling Mediated by Fungal Arginine Attenuator Peptides That Appears Independent of the Charging Status of Arginyl-tRNAs*

The Arg attenuator peptide (AAP) is an evolutionarily conserved peptide involved in Arg-specific negative translational control. It is encoded as an upstream open reading frame (uORF) in fungal mRNAs specifying the small subunit of Arg-specific carbamoyl phosphate synthetase. We examined the functions of theSaccharomyces cerevisiae CPA1 and Neurospora crassa arg-2 AAPs using translation extracts from S. cerevisiae, N. crassa, and wheat germ. Synthetic RNA containing AAP and firefly luciferase (LUC) sequences were used to program translation; analyses of LUC activity indicated that the AAPs conferred Arg-specific negative regulation in each system. The AAPs functioned either as uORFs or fused in-frame at the N terminus of LUC. Mutant AAPs lacking function in vivo did not functionin vitro. Therefore, trans-acting factors conferring AAP-mediated regulation are in both fungal and plant systems. Analyses of ribosome stalling in the fungal extracts by primer extension inhibition (toeprint) assays showed that these AAPs acted similarly to stall ribosomes in the region immediately distal to the AAP coding region in response to Arg. The regulatory effect increased as the Arg concentration increased; all of the arginyl-tRNAs examined appeared maximally charged at low Arg concentrations. Therefore, AAP-mediated Arg-specific regulation appeared independent of the charging status of arginyl-tRNA.

The Arg attenuator peptide (AAP) is an evolutionarily conserved peptide involved in Arg-specific negative translational control. It is encoded as an upstream open reading frame (uORF) in fungal mRNAs specifying the small subunit of Arg-specific carbamoyl phosphate synthetase. We examined the functions of the Saccharomyces cerevisiae CPA1 and Neurospora crassa arg-2 AAPs using translation extracts from S. cerevisiae, N. crassa, and wheat germ. Synthetic RNA containing AAP and firefly luciferase (LUC) sequences were used to program translation; analyses of LUC activity indicated that the AAPs conferred Arg-specific negative regulation in each system. The AAPs functioned either as uORFs or fused in-frame at the N terminus of LUC. Mutant AAPs lacking function in vivo did not function in vitro. Therefore, trans-acting factors conferring AAP-mediated regulation are in both fungal and plant systems. Analyses of ribosome stalling in the fungal extracts by primer extension inhibition (toeprint) assays showed that these AAPs acted similarly to stall ribosomes in the region immediately distal to the AAP coding region in response to Arg. The regulatory effect increased as the Arg concentration increased; all of the arginyl-tRNAs examined appeared maximally charged at low Arg concentrations. Therefore, AAP-mediated Arg-specific regulation appeared independent of the charging status of arginyl-tRNA.
The evidence for a role of CPA1 uORF translation in regulation is based on mutational studies (see Ref. 14, and references therein) and a variety of regulatory models are consistent with the existing data concerning the CPA1 uORF (14). For arg-2, mutational studies have been combined with direct biochemical studies. Addition of Arg to growing cells causes a rapid decrease in the rate of ARG-2 polypeptide synthesis and a decrease in the association of the arg-2 mRNA with ribosomes (7). Experiments with reporter genes show that translation of the wild-type arg-2 uORF is critical for Arg-specific translational control. mRNAs in which uORF translation is eliminated or in which the evolutionarily conserved peptide sequence is altered no longer show decreased association with ribosomes when Arg is added to cells (8,9).
Further insight into the mechanism of Arg-specific translational attenuation mediated by the arg-2 uORF-encoded peptide was gained using cap-, poly(A)-, and amino acid-dependent translation extracts derived from N. crassa in which regulation is reconstituted (15). Using reaction mixtures containing low or high Arg concentrations and a primer extension inhibition (toeprint) assay to map the positions of ribosomes on capped and polyadenylated synthetic RNA templates, high Arg concentrations are observed to cause ribosome stalling with the wild-type uORF termination codon at the ribosomal A site (16). Since reinitiation following uORF translation does not appear to be efficient in this case, these data provide the basis for a regulatory model in which the Arg-stalled ribosomes prevent trailing scanning ribosomes from reaching the downstream start codon (16,17).
Based on its cis-acting ability to repress the translation of downstream RNA sequences, the arg-2 uORF-encoded peptide was named the Arg attenuator peptide (AAP). Dissection of sequences outside of the 24-amino acid coding region established that little else of the original arg-2 mRNA is required for the AAP's regulatory function (18). Neither deletion of the intercistronic sequences nor changing the UAA termination codon to UGA or UAG alters the AAP's regulatory capacity. Furthermore, direct fusion of the AAP at the N terminus of a polypeptide results in Arg-specific stalling of ribosomes involved in elongation in the region immediately downstream of the AAP, indicating that the termination codon is dispensable for stalling.
Here we examined the generality of AAP-mediated regulation by studying the function of the CPA1 and arg-2 AAPs in translation extracts derived from S. cerevisiae, N. crassa, and wheat germ. Both CPA1 and arg-2 AAPs mediated translational attenuation in each of these systems as determined by LUC assays. Thus, factors that permit AAP-mediated translational attenuation can be found in plant as well as fungal systems. In the S. cerevisiae and N. crassa translation extracts, primer extension inhibition (toeprint) assays indicated that Arg-specific translational attenuation was associated with the stalling of ribosomes after AAP translation. The level of charged Arg-tRNA did not appear to be responsible for effecting Arg-specific control because the tRNA was fully charged even at low Arg concentrations. In contrast, all other well understood examples of translational regulation of amino acid biosynthetic genes in eukaryotes and prokaryotes that are mediated by uORFs respond to the level of tRNA charging (19,20). Thus, AAP-mediated ribosome stalling appears to be an evolutionarily conserved cis-acting control mechanism that regulates the expression of a fungal Arg-biosynthetic gene in response to Arg independent of the level of charged tRNA.

EXPERIMENTAL PROCEDURES
Templates for RNA Synthesis-Linearized plasmid templates were designed to produce capped and polyadenylated synthetic RNAs encoding firefly LUC. The first type of RNA was designed to contain the entire S. cerevisiae CPA1 uORF and intercistronic region in its 5Ј-leader (pAG101; Fig. 1B, Table I). The second type of RNA was designed to have the CPA1 uORF coding sequence fused directly in-frame with the LUC open reading frame (pAG102; Fig. 1C, Table I). Also constructed were mutant variants of each type containing either an Asp to Asn codon change at codon 13 of the CPA1 uORF (pAG103 and pAG104) or a Met to Leu codon change (AUG 3 UUG) at the predicted CPA1 uORF translation initiation codon (pAG105) by using PCR-based procedures (9). PCR products were placed into the pHLUCϩNFS4 vector (15). Primers for PCR reactions were: AG1 (5Ј-TGTTGAAGATCTACCCTT-TTTGCAGATTTG-3Ј), which includes a 5Ј-BglII site, used for pAG101, pAG102, pAG103, and pAG104; AG3 (5Ј-ATCTGACCATGGTTGAAAT-ATTTTTAGGAGTGGTT-3Ј), which includes a 3Ј-NcoI site, used for pAG101, pAG103, and pAG105; AG4 (5Ј-ATAGATGGTGACCTGGTG-GGAGCTAGTTTTCCA-3Ј), which includes a 3Ј-BstEII site, used for pAG102, pAG104, and pAG106; AG5 (5Ј-CAGATATGTAGTTTTGGCA-GG-3Ј), which contains the Asp to Asn codon change at codon 13 of the CPA1 uORF, used for pAG103 and pAG104; and AG6 (5Ј-TGTTGAAG-ATCTACCCTTTTTGCAGATTTGAAATAAAAAAAACATTATTTGTTT-AGCTTAT-3Ј), which contains a 5Ј-BglII site and changes the predicted uORF translation initiation codon, used for pAG105 and pAG106. Corresponding templates for the synthesis of RNA containing the N. crassa arg-2 AAP in the 5Ј-leader region (Table I) were described previously (15,16,18), as was the template used to produce capped and adenylated synthetic mRNA encoding sea pansy LUC to serve as an internal control for translation reactions (18).
Plasmid DNA templates were purified by equilibrium centrifugation or by using a plasmid purification kit from Qiagen; capped, polyadenylated RNA was synthesized with T7 RNA polymerase from EcoRIlinearized plasmid DNA templates, and the yield of RNA was quantified (15).
Cell-free Translation and Primer Extension Inhibition (Toeprint) Analyses-The preparation of translation extracts was as described (21) from S. cerevisiae strain YAS1874 (MATa MAK10::URA3 PEP4::HIS3 prb1 prc1 ade2 leu2 trp1 his3 ura3) (22) with two modifications: buffer A was pH 7.6 instead of pH 7.4, and extracts were treated with micrococcal nuclease immediately after recovery from the Sephadex G-25 column, prior to freezing and storage. Nuclease-treated yeast extracts were used because nuclease treatment did not significantly affect amino acid-dependence or Arg-specific regulation under our assay conditions, but in initial comparative studies greatly increased the absolute level of reporter RNA translation and yielded superior toeprints (data not shown). The preparation of translation extracts from N. crassa (with no nuclease treatment) was as described (15).
The reaction conditions for in vitro translation using S. cerevisiae and N. crassa extracts were essentially as described previously (15,21). For translation in S. cerevisiae extracts, the final concentrations of K ϩ and Mg 2ϩ were 230 and 3.4 mM, respectively. Translation reaction conditions using nuclease-treated wheat germ extracts (Promega) were essentially those specified by the supplier, except that, to achieve maximum activity, K ϩ and Mg 2ϩ final concentrations were adjusted to 100 and 2.1 mM, respectively. All reaction mixtures were incubated at 25°C; for LUC assays, translation was halted by freezing in liquid nitrogen after 30 min of incubation, and 5-l aliquots of the ice-thawed mixtures were used for analysis (15,18).
A wide range of conditions were examined to find those that were optimal for toeprinting in S. cerevisiae-derived reactions, including pH, heat pretreatment, reaction temperature, and Mg 2ϩ concentration (23). These were similar to those earlier determined to be optimal for N. crassa-derived reactions (data not shown). Therefore, the toeprint assays of both S. cerevisiae-and N. crassa-derived reaction mixtures were accomplished after incubation as described in the text using primer ZW4 and the previously established method (16). All toeprint data shown are representative of multiple experiments.
Measurement of tRNA Aminoacylation-The assays for tRNA aminoacylation were adapted from a previously described procedure (24). Translation reaction mixtures with total volumes of 100 l (S. cerevisiae and N. crassa) or 60 l (wheat germ) containing different concentrations of Arg were incubated for 10 min. Then aliquots of 90 l (S. cerevisiae and N. crassa) or 54 l (wheat germ) were removed (the remainder of translation reaction mixtures were incubated to the 30min time point, then used for LUC assays) and immediately added to ice-cold tubes containing a mixture of 300 l of phenol (pH 4.5) and 200 l of sodium acetate (pH 4.5). Tubes were vortexed for 60 s and then centrifuged for 20 min. The aqueous layers were transferred to new tubes and mixed with 2.5 volumes of ethanol. Tubes were frozen at Ϫ80°C for at least 15 min, and then the total nucleic acids were recovered by centrifugation for 20 min. The nucleic acid pellet was dissolved in 20 l of 10 mM sodium acetate, 1 mM EDTA (pH 4.5). An aliquot (2.0 l) was used to measure A 260 ; immediately prior to gel electrophoresis, nucleic acids were adjusted to a final concentration of 2.5 g/l (assuming 40 g of nucleic acid/A 260 ) by the addition of acid gel loading buffer (24).
The level of tRNA charging initially present in extracts at time T 0 was determined by processing as described above of 50 l of extract (S. cerevisiae and N. crassa) or 30 l of extract (wheat germ) with no addition of other reaction mixture components (e.g. additional salt, synthetic mRNA, and amino acids in all three cases, plus an energyregenerating system for S. cerevisiae and N. crassa). As an additional control, tRNAs in T 0 extracts were deacylated by alkali treatment. First, aliquots of T 0 extracts (50 or 30 l) were mixed with 250 l of 0.2 M Tris-HCl (pH 8.0) and extracted with 300 l of phenol (pH 8.0). Nucleic acids in the aqueous phase were precipitated with salt and ethanol; the precipitates were dissolved in 100 l of 0.1 M Tris-HCl (pH 8.8) and incubated at 37°C for 20 min to deacylate the tRNAs. After another ethanol precipitation, the pellet was dissolved in 20 l of 10 mM sodium acetate, 1 mM EDTA (pH 4.5) prior to dilution with acid gel loading buffer.
The procedures for acid/urea gel electrophoresis, electrophoretic transfer, and Northern blot hybridization to identify charged and uncharged tRNAs were essentially as described (24), except that denatured salmon sperm DNA was not included in the prehybridization and hybridization solutions, and the membranes after hybridization were exposed to screens of a Molecular Dynamics PhosphorImager for ap-

AAP-mediated Arg-specific Translational Attenuation in
Three Cell-free Translation Systems-We examined Arg-specific regulation mediated by the S. cerevisiae and N. crassa AAPs in translation extracts derived from S. cerevisiae, N. crassa and wheat germ. Capped and polyadenylated synthetic RNAs were synthesized from templates in which the S. cerevisiae CPA1 AAP or the N. crassa arg-2 AAP were placed upstream of firefly LUC, either as uORFs or as in-frame Nterminal extensions (Fig. 1, B and C, Table I). Equal amounts of each RNA were used to program translation extracts. As an internal control, a second capped and polyadenylated synthetic RNA that encoded sea pansy LUC (which lacked fungal regulatory sequences) was also added to the extracts (18). For each of the extracts used, the addition of 10 M each of the 20 amino acids to reaction mixtures was sufficient for near maximal translation of LUC (data not shown). Additional Arg could be added to translation extracts with relatively slight effects on protein synthesis from RNA templates lacking arg-2 regulatory sequences (e.g. sea pansy LUC, Fig. 5B).
Comparisons of the translation of firefly LUC in reaction mixtures supplemented with low (10 M) or high (2 mM) Arg showed that the wild-type CPA1 AAP, when present as a uORF (Fig. 1B), reduced the translation of LUC when the concentration of Arg was high ( Table I). Translation of LUC from RNA containing the wild-type N. crassa arg-2 AAP was also subject to Arg-specific negative regulation in each extract ( Table I).
Introduction of the D13N mutation in the S. cerevisiae AAP coding region (Fig. 1B) or the corresponding D12N mutation into the N. crassa coding region eliminated this regulatory effect in all cases (Table I).
The wild-type CPA1 and arg-2 AAPs, when fused directly to LUC as N-terminal extensions (Fig. 1C), also functioned to regulate translation in all three extracts (Table I). The LUC polypeptide produced appeared to initiate at the AAP start codon because changing this codon from AUG to UUG resulted in unregulated and substantially reduced (50-fold in N. crassa, 30-fold in S. cerevisiae) LUC synthesis (data not shown). The CPA1 AAP D13N mutation and the arg-2 AAP D12N mutation eliminated the regulatory effect of Arg in all three systems, showing the strong dependence of Arg regulation on the sequence of the AAP peptide and the lack of necessity of a uORF termination codon for regulation.
Ribosomal Stalling in High Arg Is Mediated by the Wild-type CPA1 and arg-2 AAPs-Toeprint assays, in which reverse transcriptase is used for primer extension in translation extracts, enables the mapping of the positions of N. crassa ribosomes on RNA (16,18). Here we applied this technique to S. cerevisiae ribosomes. To our knowledge, this is the first use of toeprinting to examine the positions of S. cerevisiae ribosomes on RNA. Primer extension from RNA templates containing the CPA1 uORF in the absence of extract yielded cDNA products predominantly corresponding to full-length extension of the primer as well as other shorter transcription products (Fig. 2, lanes 8 and  15). The shorter products are produced in relatively much lower quantities when the RNA is reverse-transcribed in buffer formulated for reverse transcription (data not shown) rather  GenBank accession no. AJ224085). B, sequences of wild-type and mutant templates in which the CPA1 AAP is encoded by a uORF. The sequence shown begins with the T7 RNA polymerase-binding site and ends within the LUC coding region (16). The 5Ј and 3Ј boundaries of the CPA1 region that was amplified by PCR are boxed. The amino acid sequences of the CPA1 AAP and the N terminus of LUC are indicated. Point mutations are shown below the wild-type sequence. The sequence for which the reverse complement was synthesized and used as primer ZW4 for toeprint analysis is indicated by a horizontal arrow below the sequence. C, sequences of wild-type and mutant templates containing CPA1 AAP-LUC fusion genes. The sequence shown begins with the T7 RNA polymerase-binding site and ends within the LUC coding region (57). The 5Ј and 3Ј boundaries of the CPA1 region that was amplified by PCR are boxed. The amino acid sequence of the N terminus of the AAP sc -LUC fusion polypeptide is indicated. Point mutations are shown below the wild-type sequence. The sequence for which the reverse complement was synthesized and used as primer ZW4 for toeprint analysis is indicated by a horizontal arrow below the sequence.
than the buffer formulated for in vitro translation necessary for these experiments. Extracts that were not programmed with RNA did not yield any of these signals (Fig. 2, lanes 7 and 16), as predicted if these represented the products obtained from priming on the synthetic RNA template. When RNA containing the CPA1 uORF in its 5Ј-leader was used to program S. cerevisiae extracts containing high Arg (500 or 2000 M, Fig. 2, lanes  2 and 3) but not low Arg (10 M , Fig. 2, lane 1), new signals were observed that corresponded to ribosomes stalled with the uORF termination codon in the ribosome A site (confirmed by high resolution mapping on other gels, data not shown). The effect of Arg to stall ribosomes at the termination codon increased when the concentration of Arg was raised from 500 M to 2000 M. Arg also caused ribosome stalling at the CPA1 uORF termination codon in N. crassa extracts (Fig. 2, compare  lanes 10 and 11 to lane 9). The CPA1 AAP D13N mutation, which eliminates regulation in vivo, (5), eliminated Arg-specific effects on toeprints in both extracts (Fig. 2, lanes 4 -6 and a For details of how the CPA1 constructs were made, see Fig. 1 legend and "Experimental Procedures"; for arg-2 constructs, see Refs. 15 and 18. Wild-type represents the wild-type CPA1 AAP or arg-2 AAP as independent uORFs with wild-type initiation contexts. AAP-LUC represents the CPA1 or arg-2 AAPs as in-frame N-terminal fusions of the AAP to LUC. D13N and D12N represent amino acid substitutions in the AAPs that abolish regulation. ⌬AUG represents an AUG to UUG mutation that eliminates the initiation codon for the uORF encoding the CPA1 AAP.

12-14)
, consistent with the loss of regulation observed by LUC assay (Table I).
Puromycin, an inhibitor of translation that releases 80 S ribosomes from RNA, would be expected to release Arg-specific signals if they arose from the stalling of ribosomes. Therefore, extracts were programmed with RNA and incubated for 15 min in low or high Arg; then puromycin was added to a final concentration of 1.3 mM (or water was added as a negative control) and incubation continued for 5 min. Extracts were then subjected to toeprint analysis. Puromycin released the Arg-specific toeprints observed in both S. cerevisiae and N. crassa extracts (data not shown). Thus, the Arg-specific signals appear to be a reversible consequence of the association of ribosomes with the RNA.
In N. crassa extracts, signals corresponding to ribosomes with the uORF and LUC initiation codons in their P-sites were observed in extracts programmed with RNA. For constructs containing the wild-type but not the D13N uORF, these signals were reduced when Arg was added. Similar results are observed for the wild-type arg-2 uORF in N. crassa extracts and are interpreted to arise as a consequence of ribosome stalling at the uORF termination codon, which decreases ribosome loading at the LUC AUG, and decreases the capacity to detect signal at the uORF AUG (16,18).
The wild-type CPA1 AAP caused additional Arg-specific effects in each extract, some common and some system-specific ( Fig. 2 and data not shown). In N. crassa extracts, an additional signal (arrowhead) approximately 30 nucleotides upstream of the stop codon appeared in high Arg. These possibly represent ribosomes queued behind ribosomes that have stalled at the uORF termination codon (16,18). In both systems, Arg caused a substantial increase in (puromycin-releasable) toeprints in the intercistronic region (asterisk, approximately 12 and 16 nucleotides downstream of the stop codon in N. crassa, and 16 nucleotides downstream of the stop codon in S. cerevisiae). In S. cerevisiae, additional Arg-regulated toeprints were observed further downstream in the intercistronic region (star and bracket). One of these signals corresponds in position to a strong signal in the intercistronic region of the RNA, which was also present in primer extension products obtained from the RNA in the absence of translation extract (Fig. 2, lanes 8 and  15, star). The physical basis for these additional bands, which may arise for reasons similar or different than those responsible for the "echo band" phenomenon, in which a ribosome located at an initiation codon can cause a primary toeprint and an additional toeprint (23), remain to be elucidated. Possibly, they could represent ribosomes or additional machinery recruited by ribosomes; alternatively, they could reflect increased secondary structure in the RNA arising as a consequence of ribosome binding.
In a manner highly similar to the wild-type CPA1 AAP, the wild-type arg-2 AAP (in these experiments placed in an improved initiation context; Ref. 16) caused ribosomes to stall at the uORF termination codon in response to Arg in both S. cerevisiae and N. crassa systems; the D12N mutation eliminated regulation (Fig. 3). Thus, the two fungal AAPs acted similarly when present as uORFs to stall ribosomes in S. cerevisiae and N. crassa systems.
While AAP-dependent regulation was observed using wheat germ extracts (Table I), in primer extension experiments, no signals indicating Arg-specific stalling were apparent (data not shown). All signals, including full-length cDNA products, were weaker in primer extension analyses using wheat germ. This failure to achieve results in the wheat germ system that were comparable to those obtained with the fungal systems is possibly attributable to the presence of an RNase H activity in wheat germ extracts (34).
The effect of Arg-specific, AAP-mediated regulation on ribosomes involved in elongation was tested using CPA1 AAP-LUC and arg-2 AAP-LUC fusions in S. cerevisiae and N. crassa

FIG. 3. Effects of the arg-2 AAP encoded as a uORF on Arg-specific regulation in translation extracts derived from N. crassa and S. cerevisiae.
Equal amounts of synthetic RNA transcripts (120 ng) were translated in reaction mixtures and analyzed by toeprinting as described in the legend to Fig. 2. The transcripts encoded either the wild-type (wt) AAP in an improved initiation context or the D12N AAP (16). Arrows indicate the positions of premature transcription termination products corresponding to ribosomes bound at AUG uORF , UAA uORF , or AUG LUC . The arrowhead indicates the position of an additional toeprint site upstream of UAA uORF observed in N. crassa extracts containing high Arg concentrations. The star indicates a signal observed from primer extension of RNA in the absence of extract (-EXT) that coincides with a toeprint signal that increases with high Arg (16). Dideoxynucleotide sequencing reactions for the wild-type arg-2 template are shown on the left; the nucleotide complementary to the dideoxynucleotide added to each sequencing reaction is indicated below the corresponding lane so that the sequence of the template can be directly deduced; the 5Ј-to-3Ј sequence reads from top to bottom. extracts. A high concentration of Arg substantially increased the intensity of a series of toeprints on the CPA1 AAP-LUC RNA in both extracts (Fig. 4, compare lanes 4 and 3, and lanes  12 and 11). The D13N mutation eliminated these Arg-specific effects on toeprints (Fig. 4, compare lanes 1 and 2, and lanes 9  and 10), as did treatment with puromycin after 15 min, as described above (Fig. 4, lanes 5, 6, 13, and 14), indicating that they resulted from an interaction of ribosomes with the RNA. In both extracts, the most 5Ј-proximal of the Arg-specific toeprints corresponded to ribosomes translating the first codon following the AAP coding sequence (Fig. 4, arrowheads). This stall site in the fusion polypeptide corresponds to the position, relative to the CPA1 AAP coding sequence, of the uORF termination codon. This toeprint site was followed by additional Arg-induced toeprints corresponding to ribosomes stalled in the downstream LUC coding region (indicated by brackets). These signals extend further downstream in S. cerevisiae extracts.
The length of the region in which ribosomes involved in elongation were stalled in response to Arg appeared to be determined by the source of the extract and not the source of the AAP. Both CPA1 and arg-2 AAPs yielded a more extended series of toeprint sites in S. cerevisiae than N. crassa ( Fig. 4; data not shown). The reasons for these extract-dependent differences in toeprinting are not known but might reflect faster translation elongation rates in S. cerevisiae-derived extracts (data not shown).
Arg-specific Regulation Appears Independent of the Charging Status of Arginyl-tRNAs-When the effect of Arg on ribosome stalling on transcripts encoding the wild-type arg-2 AAP-LUC fusion was examined in N. crassa translation extracts, the amount of stalling increased as the concentration of Arg increased (Fig. 5A, lanes 1-5). Similar effects were observed with the CPA1 AAP-Luc fusion in N. crassa extracts and with both fusions in S. cerevisiae extracts (data not shown). The S. cerevisiae D13N and N. crassa D12N mutants did not show stalling at any Arg concentration in either extract ( Fig. 4; Fig. 5A, lanes 6 -8; data not shown). Consistent with the observed increase in stalling of ribosomes on RNA containing the wild-type AAP-LUC fusion, the magnitude of Arg-specific regulation increased in S. cerevisiae, N. crassa, and wheat germ extracts as the concentration of added Arg was increased from 10 M to 5 mM (Fig. 5B) as determined by LUC assay. It should be noted that, while the precision of measurements in a given experiment is high, the absolute magnitude of regulation by Arg differs between extract preparations and experiments (e.g. the extracts used in the experiment shown in Table I showed a lower magnitude effect than those used in Fig. 5B). Nonetheless, in multiple experiments using any of the amino acid-dependent fungal extracts prepared in our laboratory to date (28 independently prepared N. crassa extracts and 12 independently prepared S. cerevisiae extracts), Arg-specific regulation is always observed.
The level of charged arginyl-tRNA might be a signal for Arg-specific regulation mediated by the AAP. Transcriptional attenuation of the amino acid biosynthetic operons in bacteria is modulated by the level of charged tRNA (20). The charging status of tRNA controls the translation of Gcn4p in yeast (19). To determine whether the levels of aminoacylation of arginyl-tRNAs change when different Arg concentrations are present in translation extracts, we adapted a method that has been successful in determining the levels of aminoacylation of FIG. 4. Effects of the CPA1 AAP as an N-terminal fusion to LUC on Arg-specific regulation in translation extracts derived from S. cerevisiae and N. crassa. Equal amounts of synthetic RNA transcripts (120 ng) were translated in reaction mixtures and analyzed by toeprinting as described in Fig. 2. The transcripts encoded either the wild-type (wt) AAP-LUC fusion or the D13N mutant AAP-LUC fusion as indicated. Puromycin (Pur) was added where indicated (ϩ), as described in the text. The arrow indicates the position of premature transcription termination products corresponding to ribosomes bound at the AAP initiation codon (AUG AAP ). The arrowhead indicates the positions of premature termination products corresponding to ribosomes stalled at the codon immediately following the last codon of the AAP in S. cerevisiae and N. crassa extracts containing a high Arg concentration. The bracket indicates the position of premature termination products corresponding to ribosomes stalled in the LUC coding region in S. cerevisiae and N. crassa extracts containing a high Arg concentration. Dideoxynucleotide sequencing reactions for the wild-type CPA1 AAP-LUC fusion template are shown on the left; the nucleotide complementary to the dideoxynucleotide added to each sequencing reaction is indicated below the corresponding lane so that the sequence of the template can be directly deduced; the 5Ј-to-3Ј sequence reads from top to bottom. tRNAs in vivo (24). Reaction mixtures were supplemented with increasing concentrations of Arg and incubated for 10 min. Then total nucleic acid was obtained under conditions in which tRNA charging is maintained and the tRNAs separated by polyacrylamide gel electrophoresis using conditions that resolve charged and uncharged tRNAs. The charging status of different tRNA species were detected by Northern blot hybridization using 32 P-labeled oligonucleotide probes complementary to the specific tRNAs of interest.
We first checked the charging status of S. cerevisiae tRNAs in yeast extracts with a positive control tRNA i Met probe. Predominantly uncharged tRNA is observed after alkali treatment of the tRNAs isolated from extracts (Fig. 6A, lane 1). tRNA is mostly uncharged in T 0 S. cerevisiae extracts, which have not been incubated and which have not been supplied with an energy regeneration system or additional amino acids (Fig. 6A,  lane 2). In contrast, in complete translation extracts containing 10, 150, 500, or 2000 M Arg and 10 M each of the other 19 amino acids that have been incubated for 10 min, the tRNA i Met is predominantly fully charged (Fig. 6A, lanes 3-6).
The arginyl-tRNAs detected with probes that should recognize three different tRNA Arg species showed the same pattern of charging as the tRNA i Met control (Fig. 6, B and C). The arginyl-tRNAs were mainly uncharged in T 0 extracts but maximally charged at even the lowest concentration of Arg added to extracts (10 M). Thus, in S. cerevisiae extracts, the charging FIG. 5. Effects of Arg concentration on Arg-specific regulation. A, effects of arg-2 AAP-LUC fusion in N. crassa translation extracts containing varying concentrations of Arg assayed by toeprinting. Equal amounts of synthetic RNA transcripts (120 ng) were translated in reaction mixtures that contained 10, 150, 500, 2000, or 5000 M Arg and a 10 M concentration of each of the other 19 amino acids. Transcripts were analyzed by toeprinting as described in the legend to Fig. 2. The transcripts encoded either the wild-type (wt) or the D12N mutant arg-2 AAP as a uORF in the 5Ј leader. The arrow indicates the position of toeprint products corresponding to ribosomes bound at the AAP initiation codon (AUG AAP ). The arrowhead indicates the position of toeprint products corresponding to ribosomes stalled at the codon immediately following the last codon of the AAP in N. crassa translation extracts containing high Arg concentrations. The bracket indicates the positions of toeprint products corresponding to ribosomes stalled in the luciferase coding region in N. crassa translation extracts containing high Arg concentrations. Dideoxynucleotide sequencing reactions for the wildtype arg-2 AAP-LUC fusion template are shown on the left; the nucleotide complementary to the dideoxynucleotide added to each sequencing reaction is indicated below the corresponding lane so that the sequence of the template can be directly deduced; the 5Ј-to-3Ј sequence reads from top to bottom. B, effects of arg-2 AAP-LUC fusion on Arg-specific regulation in translation extracts derived from S. cerevisiae, N. crassa, and wheat germ assayed by measuring luciferase enzyme activity. Equal amounts (12 ng) of arg-2 AAP-LUC fusion RNA were translated in S. cerevisiae, N. crassa, and wheat germ extracts containing 10, 150, 500, 2000, or 5000 M Arg and a 10 M concentration of each of the other 19 amino acids. Mean values and standard deviations from measuring the firefly luciferase enzyme activity in two independent translation reactions are given. Activities of sea pansy luciferase translated from a second RNA encoding this enzyme (an RNA lacking AAP regulatory sequences that was included as an internal control in each reaction mixture) are indicated below the corresponding firefly luciferase activities. status of arginyl-tRNAs did not appear to change in response to levels of Arg supplement that result in Arg-specific translational regulation.
Similar studies were attempted using N. crassa extracts. Both tRNA i Met and tRNA GAA Phe were maximally charged under normal translation conditions (data not shown). However, we were unable to detect N. crassa tRNA Arg with S. cerevisiae probes, and lacking N. crassa tRNA Arg sequences to design specific probes, we were unable to determine the charging status of these tRNAs.
Since AAP-mediated Arg-specific regulation was observed in wheat germ extracts (Fig. 5B), we analyzed the charging status of methionyl-and arginyl-tRNAs in these extracts (Fig. 6, D-F). The results were similar to those obtained with yeast extracts except that wheat germ tRNAs were already charged in T 0 extracts. The reason for this difference between wheat germ and yeast extracts was not determined, but it might reflect the presence of a high level of ATP in T 0 wheat germ extracts, since ATP and an ATP regenerating system are present in T 0 wheat germ extracts but not T 0 yeast extracts. DISCUSSION The S. cerevisiae Arg biosynthetic gene CPA1 contains a cis-acting control region functionally analogous to a bacterial operator in repressing gene expression in response to Arg (35). Regulation is indicated to act at the level of translation because mutations causing constitutive expression affect the transla-tion of a peptide encoded in the 5Ј-leader of the CPA1 mRNA (5) that is evolutionarily conserved (Fig. 1A). We examined the role of this AAP in Arg-specific regulation by programming translation extracts from S. cerevisiae, N. crassa, and wheat germ with mRNA containing the AAP and firefly LUC reporter sequences. Using fungal extracts, in which the movement of ribosomes could be examined by primer extension inhibition, the wild-type but not mutant CPA1 and arg-2 AAPs acted similarly to stall the movement of ribosomes immediately after AAP translation. Regulation did not appear to be a response to the level of charged Arg-tRNA. The observation that tRNAs were maximally charged in extracts provided with 10 M exogenously supplied amino acids is consistent with the observation that this amount of supplement is sufficient for near maximal translational activity and consistent with the observed K m of 1.5 M for Arg of the purified S. cerevisiae arginyl-tRNA synthetase (36).
The absence of an apparent role for the level of charged tRNA in a case of translational regulation of an amino acid biosynthetic gene is unprecedented. tRNA charging is important for the transcriptional attenuation of amino acid biosynthetic genes in prokaryotes, in which lack of specific charged tRNAs causes critical stalls in the translation of upstream leader peptides (20), and in the translational regulation of S. cerevisiae GCN4 through the GCN2-encoded kinase, which is activated by binding to uncharged tRNA (37). However, these latter control mechanisms that respond to tRNA charging are designed to respond to amino acid limitation. CPA1 in fact responds to Arg limitation through a GCN4-mediated process (38), and N. crassa arg-2 responds to amino acid limitation through a cpc-1 mediated process (9,39,40). cpc-1 is the homolog of GCN4 (41,42), and the translation of its mRNA is also regulated by amino acid limitation (7). The available evidence concerning the signal for the CPC-1-mediated response to amino acid limitation in N. crassa indicates that it is uncharged tRNA (42), as it is for the Gcn4p-mediated response in yeast. Strikingly, N. crassa contains a close homolog of GCN2 known as cpc-3, and cpc-3 mutants have phenotypes similar to gcn2 mutants (43). Thus, it appears that CPA1 and arg-2 share both a conserved mechanism to respond to amino acid limitation (through GCN4/GCN2 and cpc-1/cpc-3, respectively) and a conserved mechanism to respond to Arg surplus (through translation of the cis-acting AAP). The response to limitation appears mediated by the level of tRNA; the response to surplus appears to be mediated differently. Because fungi store large amounts of Arg in the vacuole (44) (the concentration of Arg in the vacuole of S. cerevisiae grown in Arg-containing medium is 430 mM (45) and a high concentration of Arg is also stored in the vacuole of N. crassa (46), it would seem logical that they possess a regulatory mechanism to modulate Arg biosynthesis in response to cytosolic concentrations of Arg far exceeding those necessary for the charging of tRNA.
The S. cerevisiae and N. crassa AAPs exerted regulatory effects on translation in plant as well as fungal systems. These data provide constraints for models of how Arg exerts its regulatory effect. Presuming that Arg, or a close metabolite, is directly responsible for regulation, then there are at least three ways that it could function to control the movement of ribosomes. High concentrations of Arg could result in modification of the translational machinery (analogous to uncharged tRNA resulting in eIF2␣ phosphorylation). This modified machinery would then be sensitive to stalling by the wild-type AAP. Wheat germ and fungal systems might share regulatory pathways (or have independently derived regulatory pathways) that enable AAP-mediated Arg-specific control to be observed in vitro. Second, Arg might not cause modification of a translational component, but instead might interact directly with the translational machinery, causing the machinery to become sensitive to AAP-mediated stalling. In addition to the well established interaction of Arg with regulatory proteins such as the E. coli Arg repressor (e.g. Ref. 47), Arg can also interact with RNA. In the case of the Tetrahymena rRNA self-splicing intron, Arg competes for GTP binding (48). The human immunodeficiency virus trans-acting responsive element RNA binds an Arg residue of Tat; it also binds the free amino acid, blocking the interaction of the RNA with Tat (49). RNA aptamers can also be selected on the basis of their Arg binding (50). There is already precedent for the direct inhibition of ribosomal peptidyl transferase activity by Arg (51). However, that inhibitory effect was elicited by either D-Arg or L-Arg, but D-Arg does not elicit AAP-mediated control in N. crassa extracts (15) or S. cerevisiae extracts (data not shown).
Finally, Arg might exert its effect by interacting directly with the AAP. The AAP-Arg complex would stall the ribosome, and thus the AAP would function as a cis-acting "argometer" from within the ribosome. This is possibly the simplest model consistent with the data available thus far, but there is yet no direct evidence supporting it relative to the other models.
That ribosomes which have translated the AAP are sensitive to stalling by Arg in extracts is clear. This effect could explain the translational response to Arg observed in vivo in N. crassa, in which Arg reduces the average number of ribosomes associated with arg-2 mRNA (7). But, in addition to reduced translation, the steady-state level of N. crassa arg-2 mRNA is also reduced by growth in Arg (7-9, 39, 52). Similarly, Arg affects the level of CPA1 transcript (53). Could there be a role for the uORF-encoded AAP in regulating the level of transcript in response to Arg in these systems, perhaps as a consequence of its function to modulate ribosome stalling? In N. crassa continuously grown in the presence of Arg, a reporter gene containing the wild-type arg-2 uORF shows a reduction in both the level of translation and the level of mRNA, as does the endogenous arg-2 gene. Introduction of the D12N mutation into the uORF of the reporter gene causes loss of regulation at both translation and mRNA levels in vivo, while the endogenous arg-2 gene remains regulated (9). The wild-type CPA1 mRNA is known to be destabilized by growth in Arg (53). One hypothesis, which remains to be tested, that could link our observations on stalling in vitro in S. cerevisiae and N. crassa systems with observations in vivo in these fungi on regulation at the level of mRNA is that ribosome stalling at the wild-type AAP termination codon in response to Arg triggers RNA destabilization. Consistent with this possibility, links between uORF termination codons and RNA stability are observed in S. cerevisiae (54,55).
In summary, translation of the evolutionarily conserved AAP in the presence of high concentrations of Arg causes ribosomes to stall. In S. cerevisiae, N. crassa, and other fungi, the AAP is encoded by a uORF in the 5Ј-leader of the transcript. The data are consistent with a model for regulation in which the AAPmediated stalling of ribosomes at the uORF termination codon in response to Arg blocks downstream initiation. Another uORF whose sequence is evolutionarily conserved, the second uORF of cytomegalovirus gpUL4 (gp48) (56), also causes ribosomes to stall after they have translated it (57,58). The existence of other uORFs whose peptide sequences are known to be important for regulation (2), such as the uORF in S-adenosylmethionine decarboxylase (59), as well as the existence of evolutionarily conserved uORFs of unknown regulatory function such as are present in transcripts specifying mammalian HER2/neu (10), bcl-2 (11,12), CCAAT/enhancer-binding pro-tein (13), and plant bZIP proteins (60), suggest that other conserved uORF-encoded peptides may prove to have special roles in regulating translation.