(Received for publication, August 2, 1996)
From the Department of Chemistry, Biochemistry, and Molecular Biology, Oregon Graduate Institute of Science & Technology, Portland, Oregon 97291-1000
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 5 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 Arg-specific 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, 19, 20, 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.
Megaprimer PCR (22) was used to obtain wild-type and
mutant arg-2 DNA fragments to which 5-BglII and
3
-XhoI sites were added. Templates for PCR reactions were
plasmids pMF11-wt and pMF11-D12N (10), which contain the wild-type
arg-2 uORF and the D12N (Asp to Asn) mutant uORF,
respectively (see Fig. 7). The arg-2 region amplified by PCR
and the nucleotide changes in mutant templates are indicated in Fig. 7.
Primers for megaprimer PCR were: ZW1 (5
-CTGAGATCTAACTTGTCTTGTCGC-3
),
which includes the BglII site; ZW2
(5
-CGCTCGAGCTTGACTTGAATGGT-3
), which includes the
XhoI site; ZL19 (5
-TTGTCGCAATCTGCCACAATGAACGGGCGCCC-3
), which puts the uORF initiation codon in a better context (
uORF); and ZL17 (5
-ATCTGCCCTTGTGAACGGGC-3
), which removes the predicted uORF
translation initiation codon (
AUG). Conditions for PCR were as described (10).
BglII- and XhoI-digested megaprimer PCR products
were gel-purified and ligated to BglII- and
XhoI-digested vectors pHLucS4 (see Fig. 8A) and
pHLuc+NFS4 (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); pHLuc+NFS4
was constructed by digestion of pHLucS4 with NcoI and XbaI, and replacement of the luciferase coding region with
the NcoI-XbaI fragment of pSPLuc+NF (Promega),
which contained the Luc+NF 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. pSPLuc+NF was linearized with EcoRI; plasmids T3LUC and T3LUCpA (26) were linearized with BamHI; and plasmids pHLucS4, pHLuc+NFS4, and their derivatives (see Fig. 8) were linearized with Ppu10I. Synthetic RNAs from pHLucS4 and pHLuc+NFS4 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 µl) to synthesize capped RNA contained: 2 µg of linearized template, transcription buffer (40 mM
Tris-HCl, pH 7.5, 6 mM MgCl2, 2 mM
spermidine, 10 mM NaCl), 10 mM dithiothreitol,
0.5 mM each of ATP, CTP, and UTP, 2 µCi of
[-32P]UTP, 0.05 mM GTP, 0.5 mM
m7G(5
)ppp(5
)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 µCi of [
-32P]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 [
-32P]UTP to the same specific
activity. Reactions (50 µl) were incubated for 1 h at 37 °C.
After incubation, 150 µl of water and 200 µl of 5 M
NH4OAc 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 µl of water. RNAs
were reprecipitated with KOAc and ethanol, washed with 70% ethanol,
and dissolved in 200 µl 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 TranslationWild-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 × 107 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 µg/ml
p-amidinophenylmethylsulfonyl fluoride (Calbiochem), 5 µg/ml each of pepstatin A, antipain, chymostatin, and leupeptin
(Sigma)). The homogenized mycelia were centrifuged in
a polycarbonate centrifuge tube for 10 min at 31,000 × 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 × 20-cm Sephadex G-25 Superfine column
that was pre-equilibrated with buffer A. Fractions containing the peak
A260 were pooled. Typical preparations yielded
6-7 ml of extract with an A260/ml = ~50.
When extracts were treated with micrococcal nuclease, they were first
adjusted to 1 mM CaCl2; 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 N2, and stored at 80 °C.
Unless otherwise indicated, standard translation reactions (20 µl) contained 10 µl of N. crassa extract; 30 mM HEPES-KOH, pH 7.6; 3.75 mM MgOAc; 150 mM KOAc; 1 mM dithiothreitol; 1 mM ATP; 0.25 mM GTP; 25 mM creatine phosphate; 3.6 µg (0.9 units) of creatine phosphokinase; 25 µM of each amino acid; 4 units of ribonuclease inhibitor; and protease inhibitors (25 µg/ml p-amidinophenylmethylsulfonyl fluoride, and 5 µg/ml each of pepstatin A, antipain, chymostatin, and leupeptin). Standard reactions were incubated at 25 °C for 30 min and stopped by freezing in liquid N2.
Standard reaction conditions were determined by analyses of the effects of different temperatures and different levels of K+ and Mg2+ on translation. Temperatures of 25-30 °C appeared optimal for cell-free translation. Varying the concentrations of Mg2+ and K+ 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.
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 [35S]methionine-labeled polypeptides was accomplished using a Molecular Dynamics PhosphorImager following SDS-polyacrylamide 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 µl of the thawed reactions (diluted with luciferase reaction buffer if necessary) to 50 µl 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.
Analyses of [35S]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. [35S]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 [35S]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 [35S]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, LUC+NF. 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 5 leader containing the uORF (wild-type uORF); (iii)
an arg-2 5
leader containing the wild-type arg-2
uORF in a better predicted initiation context (
uORF); (iv) the
wild-type arg-2 5
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
uORF); and (vi) a mutated
arg-2 5
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 acid-independent, 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 µM and the addition of the other 19 amino acids to final concentrations of 10 µM were sufficient for near maximal translation of luciferase (Fig. 6), enabling testing of the effects of adding 10 or 500 µM 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 LUC+NF (Promega)
has been substituted for LUC (Fig. 8D). In N. crassa extracts, RNAs containing the LUC+NF polypeptide coding
region instead of LUC produced similar yields of luciferase activity. The wild-type uORF construct, the 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
uORF), showed reduced translation of luciferase compared with the D12N mutation in the original uORF initiation context but still showed no
Arg-specific 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 acid-soluble
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 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 µM of Arg or 500 µM of each of these compounds. The addition of 150 µM 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.
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 synthetase 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 cell-free 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 sequence-independent 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, 9, 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 µM or 150 µM 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 L-arginine. 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.
We thank Zongli Luo for help preparing the uORF mutations, Salvador Tarun, Jr., and Alan Sachs for providing the S. cerevisiae translation mixtures and for advice, and Charles Yanofsky for critical reading of the manuscript.