INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Tryptophanase is a multifunctional enzyme that degrades L-tryptophan to indole, pyruvate, and ammonia (27) by a -elimination reaction (34). Bacteria with tryptophanase activity can utilize tryptophan as a source of carbon, nitrogen, and energy (23). In addition, since the -elimination reaction is reversible, tryptophanase can synthesize L-tryptophan from indole and L-serine, L-cysteine, or pyruvate and ammonia (34, 53).

Transcription of the tna operon of E. coli has been shown to be regulated by catabolite repression of transcription initiation (1, 39, 40) and by tryptophan-induced transcription antitermination (46). The tryptophanase (tna) operon has been cloned and sequenced from Escherichia coli (10), Proteus vulgaris (24), Enterobacter aerogenes (26), Symbiobacterium thermophilum (22), and the pathogen Haemophilus influenzae type b (30). The tna operon of E. coli contains two major structural genes, the promoter proximal gene, tnaA, encoding tryptophanase, and a distal gene, tnaB, encoding a low-affinity, high-capacity, tryptophan permease (10, 12). tnaA is preceded by a transcribed regulatory leader region containing a short open reading frame, tnaC, specifying a 24-residue leader peptide. Between tnaC and tnaA there is a ca. 200-bp spacer region containing several transcription pause sites. When cells grow in the absence of inducer, tryptophan, transcription is subject to Rho-dependent transcription termination (45) at these pause sites. In the presence of inducer, termination at these sites is prevented and expression is elevated 15- to 100-fold (46). Induction requires translation of tnaC, which encodes a peptide with a single Trp residue at position 12. Tryptophan induction is not observed when the tnaC start codon is replaced by a stop codon, or when the Trp codon at position 12 is replaced by codons specifying other amino acids (13, 17, 47). Additional residues in the TnaC peptide are also essential for induction, particularly several residues near the Trp residue (15).

Several classes of tna operon constitutive mutants have been isolated; these show elevated expression of the operon in the absence of added tryptophan (46). Many constitutive mutants have single base pair changes in a 9-bp sequence at the distal end of tnaC (46) which has homology to boxA of bacteriophage  (14). boxA is critical for proper antitermination, not termination, in phage  (9) and other systems (21, 50). Deletion of a putative Rho utilization (rut) site located immediately following the tnaC UGA stop codon also results in constitutive expression of the operon (17). Because the boxA-like sequence in the tna operon does not behave like a typical boxA, it is conceivable that recognition of both the boxA-like sequence and the presumed rut site are required for Rho-dependent transcription termination in this operon. A third class of constitutive mutants contains +1 or 1 frameshift mutations in tnaC that allow translation to proceed well beyond the presumed rut site that follows the tnaC stop codon (17). Continued translation presumably interferes with Rho action. However, it is unlikely that tryptophan induction involves frameshifting, because stop codons introduced in the +1 and 1 reading frames between Trp codon 12 and the tnaC stop codon do not affect either basal level expression or tryptophan induction (15). Similarly, introducing stop codons in the +1 or 1 reading frame immediately following tnaC does not interfere with the effects of induction (25).

The precise mechanism by which added tryptophan induces tna operon expression is not known. Previous studies indicated that no unique regulatory factor other than TnaC is needed for induction (15, 16). Specifically, a plasmid containing a foreign promoter, the E. coli tna leader regulatory region, and a lacZ translational fusion to the first 20 codons of tnaA responded to tryptophan induction when introduced into two bacterial species that do not produce tryptophanase (15). Thus, if additional factors are required for induction, they are likely to be components common to many bacterial species, including those that lack a tna operon (15). All tests of TnaC trans-activation have given negative results (46).

A hypothesis consistent with all the observations made to date is that in the presence of inducer, the nascent TnaC peptide acts in cis on the ribosome translating tnaC and inhibits its release at the tnaC stop codon. The inhibited ribosome would interfere with Rho's binding to the tna transcript, thereby preventing transcription termination. If ribosome release or stalling were essential in the regulation of tna operon expression, then peptide chain release factors and the ribosome release factor could be influential in setting the basal or induced level of expression of the operon. In E. coli, two codon-specific peptide chain release factors (RF1 and RF2) direct termination of protein synthesis; RF1 recognizes UAG and UAA stop codons (54), whereas RF2 acts at the UGA and UAA stop codons (6). Unlike other prokaryotic RF2s, E. coli RF2 terminates translation weakly at UGA and UAA stop codons (33, 48). The third release factor, RF3, enhances the activity of RF1 and RF2 and lacks nonsense codon specificity (18, 31).

In previous studies performed to examine models invoking ribosome release or stalling, we designed and tested a construct that would not be subject to Rho regulation. This construct, tnaC-UGA-'lacZ, lacks the tnaC-tnaA spacer region within which Rho-dependent termination occurs (28). Its expression is dependent on translation of the UGA codon and synthesis of a TnaC-'LacZ fusion protein. Using this construct and various nonsense suppressors, we showed that addition of inducer inhibited -galactosidase (-Gal) formation and that this inhibition was dependent upon the presence of Trp residue 12 in the TnaC portion of the fusion protein (28). We found that inactivation of the structural gene for RF3 (55), increased -Gal production 30-fold, and this increase was also reduced by the presence of tryptophan (28). In a parallel study with deletion derivatives of the tna operon of P. vulgaris, it was observed that a deletion that places the Shine-Dalgarno region for tnaA (in tnaA'-'lacZ fusion) near the tnaC stop codon results in tryptophan inhibition of tnaA'-'lacZ expression (25).

In this report, we extend our examination of the role of the tnaC stop codon by replacing the natural UGA stop codon with UAG and UAA. We show that the presence of UAG or UAA leads to a reduction of both basal and induced expression with tnaC-stop codon-tnaA'-'lacZ constructs containing the intact tnaC-tnaA spacer region. We also show that a temperature-sensitive mutation altering RF1 increases basal level expression in strains with UAG or UAA as the tnaC stop codon, but not in a strain with UGA. The induced levels of expression in these same strains are indistinguishable. We also examine the role of the three stop codons in a tnaC-stop codon-'lacZ fusion construct and show that tryptophan addition inhibits reading past the stop codon; however, this inhibition is not as pronounced when UAG or UAA replaces UGA. An RF1 mutation increased expression of the UAG construct in the presence and absence of tryptophan. These findings support the hypothesis that in the presence of tryptophan, TnaC peptide inhibition of ribosome release at the tnaC stop codon may be the crucial event that regulates Rho-dependent termination in the tna operon leader region.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. The E. coli strains and the plasmids used in this study are listed in Table 1. Strains VK700, VK800, and VK900 and their derivatives are all single lysogens carrying RS45 (44) with various inserts. To prepare these strains, the tnaC-UA(G/A)-'lacZ and the tnaC-UAG-tnaA'-'lacZ fusions from pRS552 (Table 1) were independently crossed into phage RS45 (44) and the recombinant phage genome was inserted into the chromosome of CY15076 (Table 1). prfA1(Ts) (37, 38) or the serU suppressors [su UAG or su UG(A/G)] (2, 28) were introduced into VK100 (28), VK700, or VK800 by P1 transduction. Mutant trpA9761 is an amber mutant of trpA in E. coli. The trpA9761 allele was introduced into appropriate strains by transduction. Plasmids were introduced into various strains by transformation (41), selecting for the appropriate antibiotic resistance marker.

Media and enzyme assay. Vogel and Bonner minimal medium (49) was used throughout. For -Gal assays (32), cultures were generally grown with shaking at 30 or 37°C in minimal medium plus 0.2% glycerol-0.05% acid-hydrolyzed casein, with or without L-tryptophan (100 µg/ml). When appropriate, media were supplemented with kanamycin (30 µg/ml), tetracycline (15 µg/ml), or ampicillin (100 µg/ml). -Gal assays were performed as described by Miller (32); -Gal activity is reported in Miller units (32). For tryptophan synthetase (TSase)  and ₂ assays, cultures were grown overnight at 37°C in the same minimal medium supplemented with 0.2% glycerol, 0.05% acid-hydrolyzed casein, and indole (4 µg/ml). Cells were collected by centrifugation, washed, and frozen. Each frozen pellet was resuspended in 0.1 M Tris buffer, pH 7.8, and disrupted by sonication. The sonication extracts were centrifuged and the cell supernatants were assayed for TSase activity in the indole plus serine to tryptophan reaction (8) in the presence and absence of excess wild-type TSase  or ₂. For a definition of the unit of specific activity, and for other assay conditions, see the work of Creighton and Yanofsky (8).

Site-directed mutagenesis. Two PCR-based methods (28) were used to introduce point mutations in the tnaC region. In the standard PCR approach, the VK1 primer, 5'-CGG AAT TCA GCT TCT GTA TTG GTA AG-3', has a sequence of nucleotides identical to the region upstream of tnaC and contains an EcoRI site at its 5' end. The mutagenic TNAC-UAG primer, 5'-GGG ATC CCC GGG AAT CTA AGG GCG GTG ATC-3', and TNAC-UAA primer, 5'-TCC CCC GGG AAT TTA AGG GCG GTG-3', contain a BamHI site and a SmaI site at their respective 5' ends. Primer pairs VK1 and TNAC-UAG or VK1 and TNAC-UAA were used with Taq polymerase (Boehringer Mannheim Co., Indianapolis, Ind.) to amplify the 353- and 347-bp PCR products, respectively. These products were individually cloned into the pCRII vector (Invitrogen Co., San Diego, Calif.), and sequences were confirmed by dideoxy sequencing (42). The 353-bp insert in the pCRII vector was digested with EcoRI and BamHI, purified by using the GENECLEAN II kit (BIO 101 Inc., La Jolla, Calif.), and subcloned into the EcoRI-, BamHI- cleaved pRS552 (44) to give the tnaC-UAG-'lacZ construct. To prepare the tnaC-UAA-'lacZ construct, the 347-bp insert in the pCRII vector was first cloned into the EcoRI-, and SmaI- cleaved sites of pBluescript (41). The resulting recombinant was digested with EcoRI and BamHI, and the insert was subcloned into the EcoRI-, BamHI-cut sites of pRS552 (44).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Basal and induced expression from tnaC-stop codon-tnaA'-'lacZ constructs in which UAG or UAA replaces UGA as the tnaC stop codon. To determine whether the identity of the tnaC stop codon influences basal or induced expression of the tna operon, constructs containing each of the three stop codons were prepared and integrated into the E. coli chromosome, and the -Gal activities of the resulting strains were determined. The natural tnaC stop codon of E. coli is UGA (Fig. 1) (10, 46). With constructs containing the UAG and UAA stop codons, basal expression of the operon was 20 and 50%, respectively, of that observed with the UGA stop codon (Table 2). These findings are consistent with evidence demonstrating that in E. coli, RF1 terminates translation more efficiently than RF2 (7, 48). Following growth with tryptophan as inducer, the same pattern was observed; induced expression was only 15 or 60%, respectively, of that obtained with the tnaC-UGA stop codon construct (Table 2). These findings establish that both basal and induced expression of the tna operon are influenced by the identity of the tnaC stop codon.

View larger version (10K): [in this window] [in a new window]   FIG. 1.   Schematic representations of the basic tna operon-'lacZ fusions employed in this study. (A) tnaC-UGA-tnaA'-'lacZ construct in strain SVS1144. The underlined rut site is part of the natural tnaC-tnaA spacer region, which also contains Rho-dependent termination sites (45). (B) tnaC-UGA-'lacZ construct in strain VK100. tnaC, its UGA stop codon, and a five-codon junction are in phase with lacZ lacking its first eight codons. The UGA stop codon (shown in boldface type) in both constructs was also changed to UAG or UAA.

Effects of the presence of a mutant RF1 on tna operon expression. Mutant allele prfA1 encodes a temperature-sensitive RF1. This allele was introduced into appropriate strains, and basal and induced expression were measured (Table 3). Since strains with prfA1 are temperature sensitive for growth, they were grown at the permissive temperature, 30°C (38). The prfA1 allele increased basal expression of the constructs with the UAG and UAA stop codons 60- and 3-fold, respectively; it reduced both basal and induced expression of the construct with the wild-type tnaC UGA stop codon by 50%. Induced expression levels of strains with the prfA1 allele and any of the three tnaC stop codons were virtually identical (Table 3). The prfA1 allele increased basal level expression of the construct in strain PDG1184 sevenfold, but there was little induced expression (Table 3).

Test for suppression of the trpA9761 amber mutation in VK801 derivatives. The high basal level expression observed in strain VK801 (tnaC-UAG-tnaA'-'lacZ/prfA1) (Table 3) could be due to specific suppression of the tnaC UAG stop codon by a resident UAG suppressor. Suppression would allow translation to continue four codons downstream from the tnaC stop codon. Previous findings argue against this possibility. First, suppression of the tnaC stop codon should at best approach expression in strain PDG1184, in which a sense codon replaces the tnaC stop codon. As shown in Table 2, strain PDG1184 does not exhibit high basal expression in the absence of tryptophan. Second, suppression of the tnaC UAG stop codon in strain VK800 (tnaC-UAG-tnaA'-'lacZ) gives basal and induced expression levels comparable to those of strain PDG1184 (Table 2).

View this table: [in this window] [in a new window]   TABLE 4.   TSase  and 2 activities of trpA prfA1(Ts) strainsa

Effects of introducing UGA stop codons in the 1 or +1 potential reading frames immediately following the tnaC UAG stop codon. Point mutations were introduced in the spacer region of the tnaC-UAG-tnaA'-'lacZ construct to test whether there is ribosomal frameshifting beyond the tnaC UAG stop codon in the prfA1 mutant background (Table 5). The resulting constructs tnaC-UAG UUUGAC and tnaC-UAG UUGACC, introduce out-of-frame UGA stop codons (shown in boldface type). In a prfA1 mutant background, any 1 or +1 frameshift would allow translation to terminate at a UGA codon immediately following the normal UAG stop codon and would presumably result in -Gal production similar to that obtained in strain PDG1184 (Table 3). In Table 5, the presence of the prfA1 allele resulted in a 5- to 10-fold increase in the basal level of -Gal activity in strains with the constructs tnaC-UAG UUUGAC (strain VK1101) and tnaC-UAG UUGACC (strain VK1201), respectively; a similar increase (18-fold) in the basal level of -Gal expression was observed with the control strain VK801 (Table 5). These findings indicate that ribosomal frameshifting beyond the tnaC UAG stop codon is not responsible for the increase in the basal level expression in a prfA1 mutant background. Note that the changes introduced following the tnaC stop codon do influence the absolute levels of basal and induced expression.

View this table: [in this window] [in a new window]   TABLE 5.   Introducing the UGA stop codon in the 1 or +1 reading frame immediately following tnaC-UAG does not appreciably alter the response to prfA1

Tryptophan inhibition of translational readthrough beyond the tnaC stop codon in constructs lacking the tnaC-tnaA spacer region. In a previous study (28) we designed and tested a construct that we believed would permit analysis of the effects of inducer in the absence of Rho-dependent termination. This construct, tnaC-UGA-'lacZ, lacks the tnaC-tnaA spacer region. It contains the tna promoter through the tnaC UGA stop codon, has an added in-phase five-codon junction, and is followed by lacZ minus its first eight codons (5) (Fig. 1). Since this construct lacks the tnaC-tnaA spacer region, Rho action should be eliminated. Our previous findings supported this expectation (28).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Studies with the tna operon of E. coli have identified features of the tna leader region that are necessary for this operon's regulation by transcriptional attenuation. In the absence of an inducer, transcription is terminated by the action of Rho factor, at one of several transcription pause sites located in the leader region preceding tnaA, the first major structural gene of the operon (45). Mutational changes that allow high-level constitutive expression of the operon in the absence of inducer have identified cell components and sequences that are necessary for transcription termination. These mutations alter Rho factor, change bases in critical regulatory sequences in the tna leader region (boxA, the rut site), or allow translation from the tnaC coding sequence to proceed in other reading frames, beyond the presumed rut site in the tna transcript. These observations and other experimental findings suggest that when cells grow without inducer, Rho factor generally binds to the leader segment of the tna transcript, moves 3' on the transcript until it interacts with a paused polymerase molecule in the leader region, and then directs the polymerase to terminate transcription.

In the presence of added tryptophan, Rho-dependent termination in the leader region of the tna operon is prevented. Induction requires translation of a 24-residue peptide coding region, tnaC, located near the 5' end of the transcript (17). Several residues of TnaC are essential for induction, suggesting that the peptide as such participates in the induction process. Some tryptophan analogs which do not appear to be incorporated into protein also function as inducers; thus, an inducer may act without being incorporated into the TnaC leader peptide (15, 16). Stop codons introduced in the +1 or 1 reading frames within tnaC, or following tnaC, do not affect either basal or induced expression, establishing that tryptophan induction does not involve frameshifting (15, 16, 25). Introducing an in-phase stop codon within the distal segment of tnaC does prevent induction (15).

How added tryptophan is used as the signal that leads to inhibition of Rho-dependent termination is a basic unanswered question. One possibility is that the TnaC peptide is modified in some manner when produced in cells growing with excess tryptophan and that the altered peptide blocks Rho action. Alternatively, in the presence of added tryptophan some cell component could interact with the TnaC peptide, altering its properties so that it blocks Rho action. Two previous studies (28, 55) and this report have focused on how the peptide may act in preventing termination. In prior analyses with a suppressed tnaC-stop codon-'lacZ fusion lacking the tnaC-tnaA spacer region, the addition of tryptophan led to inhibition of translation beyond the tnaC stop codon (28). Tryptophan addition also reversed the increased expression resulting from inactivation of RF3 (28). In related studies with the tna operon of P. vulgaris, it was shown that added tryptophan inhibited expression of a tnaA'-'lacZ translational fusion in which the tnaA ribosome binding site was placed close to the tnaC stop codon (25). These effects of inducer were not observed when the critical tryptophan codon in the leader peptide coding region was replaced by some other codon (28). The most straightforward interpretation of these observations is that in the presence of tryptophan, the TnaC leader peptide acts in cis on the translating ribosome, preventing its release at the tnaC stop codon. The stalled ribosome would block Rho's access to the transcript, thereby inhibiting termination.

In the present study we examined the effect of changing the natural tnaC UGA stop codon of E. coli to the other stop codons, UAG and UAA. We also analyzed the effects of introducing a temperature-sensitive mutant allele, prfA1, that alters RF1. With a lacZ reporter construct that allows Rho-dependent termination, changing the natural UGA stop codon to UAG lowered basal and induced expression of the operon by ca. 80%. Changing the UAA stop codon to UGA reduced basal and induced expression by 50%. The reduced expression observed with the UAG stop codon relative to the UGA stop codon could be explained most simply if RF1 dissociates the translating ribosome from the tnaC UAG stop codon more rapidly than does RF2 at the UGA stop codon. More rapid release could allow more rapid binding of Rho factor. This interpretation is consistent with the conclusion that in E. coli RF1 terminates translation more efficiently than RF2 (33, 48). Support for this interpretation is also provided by the observation that a mutant allele specifying a temperature-sensitive RF1, when examined at the permissive temperature, led to a 60-fold increase in basal level expression when the tnaC stop codon was UAG (Table 3). In fact, expression was increased only twofold further by the addition of tryptophan.

It is also conceivable that in the presence of the mutant RF1, the nucleotide sequence including the UAG stop codon permits efficient translational frameshifting and reading into the presumed rut site. However, introduction of a UGA stop codon in the 1 or +1 reading frame immediately following the tnaC UAG stop codon did not eliminate the increased basal level expression of the tnaC-UAG construct in a prfA1 mutant background. Furthermore, our findings with an in-phase tnaC-UAG-5 codon junction-'lacZ construct (Table 6) that requires suppression of the UAG stop codon for -galactosidase production also argue against this possibility. Thus, it appears that the rate of ribosome release at the tnaC stop codon (and not ribosomal frameshifting beyond the tnaC UAG stop codon) may be the key event that determines whether Rho factor will terminate transcription in the leader region of the operon.

In an early study with the tnaC-UGA-5 codon junction-'lacZ construct, we observed that under a variety of conditions allowing in-phase reading of the tnaC UGA stop codon, addition of tryptophan to the culture medium led to ca. 80% inhibition of -Gal production (28). In this study, the UGA stop codon was changed to UAG or UAA, and the resulting constructs were tested in the presence of appropriate suppressors. We found that in all but one strain (containing a Trp-inserting UAG suppressor) tryptophan addition had a 30 to 60% inhibitory effect on -Gal production (Table 6). Introduction of the prfA1 alteration resulted in a threefold increase in translation of the UAG stop codon (Table 7). In this case, the altered RF1 clearly allows increased in-phase translation. Tryptophan addition to this strain led to 50% inhibition of -Gal production (Table 7). These findings demonstrate that stop codon identity does influence the ability of tryptophan to inhibit ribosome movement beyond the suppressed tnaC stop codon.

Inhibition of ribosome function by a leader peptide has been documented in other systems. In prokaryotes, the five-residue leader peptide, MVKTD, encoded by the cat leader sequence (19, 29), and the eight-residue peptide, MSTSKNAD, encoded by the cmlA leader sequence, act in cis to block movement of the translating ribosome (29) by inhibiting its peptidyltransferase activity (20, 29). In both cases, blocking the translating ribosome allows downstream translation of a structural gene conferring drug resistance. In eukaryotes, the 24-residue upstream open reading frame (uORF) encoded upstream of the arg-2 gene of Neurospora crassa (51, 52) and the 22-residue uORF2 encoded by the gp48 gene of cytomegalovirus (3, 4, 11) have been reported to act in cis to block release of their translating ribosomes. In these cases, stalling of the translating ribosome at the respective uORF termination codon appears to inhibit translation of a downstream ORF. Additional analyses focused on the TnaC peptide of the tna operon should reveal how TnaC acts to increase this operon's expression.